JP2004187562A - Dna microarray data analyzing method, dna microarray data analyzer, program, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a DNA microarray data analyzing method allowed for analysis of mass gene data in high speed and accuracy by application of MT system method, a DNA microarray data analyzer, a relevant program, and a relevant recording medium. <P>SOLUTION: The DNA microarray data analyzing method involves narrowing down genes to be analyzed based on the assay data of DNA microarrays using the MT system method. When the MT system method is applied for analyzing the assay data for the DNA microarrays, the test data and specimen data of the DNA microarrays serve as input data and the prediction of the state of the specimen or the change predicted in the future for the state serves as the result of analysis, and from the beginning, in the process of grouping patients, the genes to be analyzed are narrowed down while judging the accuracy. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a DNA microarray data analysis method, a DNA microarray data analysis device, a program, and a recording medium, and in particular, by applying a Mahalanobis Taguchi system (hereinafter, referred to as “MT system method” or simply “MT system”). The present invention relates to a DNA microarray data analysis method, a DNA microarray data analysis device, a program, and a recording medium that can analyze a large amount of gene data at high speed and with high accuracy.
[0002]
[Prior art]
The DNA microarray is a new technology developed in recent years, and various attempts have been reported on a method of analyzing the data (Non-Patent Documents 1, 2, 4 to 7, 8 to 12 and the like). See :), not yet established. Hereinafter, a conventional DNA microarray data analysis method will be described.
[0003]
DNA microarrays can simultaneously measure the expression of hundreds to thousands or tens of thousands of genes at once. However, the advantage of the large number of measurements that must be at the same time also makes data analysis difficult. In addition, when genes whose functions and the like have not been determined are mounted on a DNA microarray, the analysis of the DNA microarray data including the genes becomes “discrimination without external standards” described later, and thus the verification of the analysis method itself becomes a problem. . Although it is easy to take data with a DNA microarray, the fact that the problem is its analysis is common knowledge among researchers.
[0004]
The center of data analysis of DNA microarrays is how to classify a large number of mounted genes based on the data. This classification is broadly divided into (1) category determination (group discovery) and (2) category assignment. The former is, for example, a case where when trying to classify genes by their expression patterns over time, it is first necessary to determine what kind of expression patterns and how many types (group discovery). In this case, there is no external criterion to determine whether the determined category is correct, and it is a so-called “unsupervised” classification. On the other hand, the latter (assignment to a category) is, for example, deciding which of the categories determined by the former the unknown gene belongs to. In this case, the criteria in the category are met or not, but verification can be performed. , So-called “supervised”.
[0005]
The determination of the category includes hierarchical clustering, a self-organizing map (SM), Gene shaving, and the like, which will be described later. The assignment to the category includes a support vector machine (SVM), discriminant analysis, and the like.
[0006]
Conventionally, a method for analyzing gene expression data in a DNA microarray has been developed as an analysis of an expression pattern accompanying the development of a DNA microarray. In other words, initially, the most developed was to comprehensively classify all genes on the array from the characteristics of the expression pattern of each gene using microarray data of cDNA (complementary DNA) such as yeast. This is a gene expression analysis for the purpose of performing the analysis (for example, see Non-Patent Document 1).
[0007]
In Non-Patent Document 1, in order to classify all the data obtained by the microarray and to estimate the functions of unknown genes contained in the data from the classes, “hierarchical clustering” (included in the above “category determination”) Is used). That is, for each gene, a vector featuring the expression state under various processing conditions is set, and the similarity between the genes to be compared is defined by Pearson's correlation coefficient, and the input value of the cluster analysis is set. (Distance measure), and in the hierarchical clustering for grouping in the next step, the feature is that “average-linkage method” is used (for example, see Non-Patent Document 3). Further, in this document, it was considered that the function of an unknown gene grouped by this method was estimated from the group of known genes in the group.
[0008]
Here, what method is used to analyze gene expression data in a DNA microarray depends on the purpose of the analysis, but even if the purpose is the same, various methods are used depending on the paper as described above. . In other words, there have been several established statistical analysis methods and pattern recognition methods, but depending on which method is combined and how, and what kind of unique techniques are added to the DNA microarray. Due to the different results, various analyzes have been attempted.
[0009]
As a DNA microarray analysis method, an analysis method (classification of cancer or classification of drug sensitivity) linked to clinical data has recently been reported (for example, see Non-Patent Document 2). This includes comprehensive analysis of the above-mentioned cDNA as a part thereof, and is a more complicated analysis method as a whole.
[0010]
In Non-Patent Document 2, a group of genes is found while reducing the dimensions, and then the patients are classified. That is, the reduction and grouping of data dimensions are performed by principal component analysis (PCA) and neural networks, and 96 genes are narrowed out of 6567 types of genes on the microarray. Subsequently, by quantifying the data of these 96 genes for each patient and performing grouping by hierarchical clustering for each patient, it was possible to accurately classify cancers that were clinically difficult to classify based on gene expression. Reporting. And this report improved the current situation that DNA microarray analysis did not have a clear difference in the conventional clinical classification, and performed the same treatment where different treatments should be originally performed. Suggested potential for great benefits.
[0011]
As described in Non-Patent Document 2, in general, in clinical data analysis using a DNA microarray, which gene is used for the purpose of analysis is not predetermined, and the expression pattern of hundreds or thousands of genes is analyzed first. It is common practice to narrow down (classification and dimensionality) which gene data to use, and then analyze the gene expression data of individual patients. In this case, since the gene narrowing is a data classification without external criteria (so-called “unsupervised”), how to verify the result is a problem. Ultimately, it is possible to verify to some extent by the results of patient grouping using the narrowed down genetic items, but actually performing narrowing down in parallel with the verification requires a huge amount of time with general analysis methods It is not realistic.
[0012]
Although there is no report on clinical data, application of a self-organizing map (SOM: Self Organization Map) (for example, see Non-Patent Document 4) using a layered neural network as “unsupervised learning” analysis of a microarray. Has been proposed (for example, see Non-Patent Document 5). This is because the above-mentioned hierarchical clustering method has disadvantages such as an increase in the amount of calculation as the number of genes increases, and the topology of the tree diagram tends to change depending on the given data set. Has been proposed as an improvement.
[0013]
In addition, attempts to apply, for example, "Gene shaving" (for example, see Non-Patent Document 6) and Correspondence analysis (CA) (for example, see Non-Patent Document 7) have been reported. However, regardless of which method is used, it is "unsupervised learning", so the validity of the analysis result cannot be verified, or even if it can be verified with clinical data, the verification itself is impractical and the final result is obtained There is a problem that can not be determined until.
[0014]
On the other hand, the above-mentioned method of analyzing the relevant gene data of a patient after gene narrowing and grouping is applicable to “supervised learning” that has external criteria in advance and classifies them into target categories based on the criteria. I do. This is applied not only to classification of patients, but also to classification of individual genes by function or the like. For example, it is proposed to adopt “supervised learning” in the above-described hierarchical clustering (for example, see Non-Patent Document 2), or to apply SVM (Support Vector Machine), and to use an analysis method by a neural network. (For example, see Non-Patent Document 8).
[0015]
As described above, various methods are currently being tested as methods for analyzing data of the DNA microarray. In particular, how to narrow down genes without the above-mentioned external reference data is becoming an even bigger problem with the recent increase in attempts to apply DNA microarrays to clinical diagnosis.
[0016]
The following is a report of microarray analysis in clinical diagnosis. As main ones, there are Non-Patent Documents 9 to 12 in addition to Non-Patent Document 2 described above.
[0017]
This non-patent document 9 reports on prediction of recurrence of cancer (NSCLC). Here, 2899 genes are characterized using a method called “Cox proportional hazards model” in advance, and then the data is obtained. And 2899 genes are used for hierarchical clustering.
[0018]
Non-Patent Document 10 reports on the prediction of the efficacy of drug therapy in breast cancer. Here, 231 genes were selected from 25,000 genes using basically the same method as Non-Patent Document 2 described above. Then, hierarchical clustering is performed based on the data of each patient.
[0019]
Non-Patent Document 11 reports on the prediction of the efficacy of drug therapy in esophageal cancer. Here, for each of the 9216 gene data, the U value of the Mann-Whitney test was statistically calculated and 52 Genes are selected, and DRS (drug response score) is calculated based on the expression status of those genes.
[0020]
Furthermore, Non-Patent Document 12 reports that a cure rate (survival rate) in DLBCL (Diffuse large B-cell lymphoma) was predicted.
[0021]
On the other hand, the MT system method is a pattern recognition method developed mainly for quality engineering, and is used for data monitoring of each device for device monitoring and preventive maintenance (for example, see Patent Document 1). In addition, although there are reports such as a prediction method of a health check by the MT system method (for example, see Non-Patent Document 13), there is no report that applies the MT system method to gene data analysis in DNA microarray analysis.
[0022]
The reason is that in the MT system method, the number of reference data (in the embodiment of the present invention, the number of individual samples) is the number of items (in the embodiment of the present invention, the number of genes on a DNA microarray plus the number of clinical data such as blood tests). Dr. Genichi Taguchi, who advocated the MT system method, has been repeatedly stated in academic societies and the like that it must be collected more than ever, and since it was established as common general knowledge for those skilled in the art, the large number of genes This is because it was considered unsuitable for the analysis of the DNA microarray, which is the most characteristic.
[0023]
Here, the number of reference data is generally encouraged to be at least twice the number of items. For example, in Non-Patent Document 14, when the number of items is 26, the number of reference data is “the lowest Mahalanobis distance can be calculated. The number of data, 52, could not be obtained. "Therefore, it is widely known in the analysis using the MT system method that the number of items is intentionally reduced to 22, for example.
[0024]
[Patent Document 1]
JP-A-2000-259222
[Non-patent document 1]
Eisen, et. al. , "Cluster analysis and display of genome-wide expression patterns.", Proc. Natl. Acad. Sci. , 1998, 95, p. 14863-14868
[Non-patent document 2]
J. Khan, et. al. , "Nature Medicine", 2001, Vol. 7, Num. 6, p. 673-679
[Non-Patent Document 3]
Sokal, et. al. , "Univ. Kans. Sci. Bull.", 1958, 38, p. 1409-1438
[Non-patent document 4]
Kohnen T .; , "Proc. IEEE 78", 1991, p. 1464-1480
[Non-Patent Document 5]
Tamayo, et. al. , "Interpreting patterns of gene expression with self-organizing maps: methods and application to hematopoietic differentiation", Proc. Natl. Acad. Sci. , 1999, 96, p. 2907-2912
[Non-Patent Document 6]
Hstie, et. al. , "Gene shaving as a method for identifying distinct sets of genes with similar expression patterns", Genome Biology, 2000, 1 (2), p. research0003.1-0003.21
[Non-Patent Document 7]
See Fellenberg, et. al. , "Correspondence analysis applied to microarray data", Proc. Natl. Acad. Sci. , 2001, 98, p. 10781-10786
[Non-Patent Document 8]
Maeda Eisaku, "Pleasant! Support Vector Machine", Information Processing, 2001, Vol. 42, No. 7, p. 676-683
[Non-Patent Document 9]
D. A. Wigle, et. al. , "Cancer Research", 2002, 62, p. 3005-3008
[Non-Patent Document 10]
L. J. van'tVeer, et. al. , "Nature", 2002, Vol. 415, p. 530-536
[Non-Patent Document 11]
C. Kihara, et. al. , "Cancer Research", 2001, 61, p. 6474-6479
[Non-Patent Document 12]
M. A. Shipp, et. al. , "Nature medicine", 2002, Vol. 8, Num. 1, p. 68-74
[Non-patent document 13]
Nakajima et al., "Predictive Evaluation and Efficiency of Health Examination Using the MT System Method," Nihon Koei Magazine, Vol. 46, No. 5, 1999
[Non-patent document 14]
"Profit Forecast Using MTS,""QualityEngineering," Vol. 10, No. 3, pp. 96-102, June 2002
[0025]
[Problems to be solved by the invention]
However, the conventional gene analysis method using a DNA microarray has a problem that it takes a lot of time to analyze a large amount of gene data covering several hundreds to tens of thousands of items.
[0026]
Specifically, for example, in the conventional method using a neural network that requires teacher data, the features of all the teacher data (data for which the determination result is clear; for example, the sensitivity of a clinically determined drug) are quantified. Because of the necessity of expression, there is a problem that a large amount of time is required for analyzing a large amount of genetic data.
[0027]
For example, in the conventional method using neural network learning and multiple regression analysis coefficient calculation, the same operation is repeated multiple times to form neural circuits and parameters in an optimal state. There is a problem that it takes a lot of time.
[0028]
Further, the conventional method using a neural network or the like has a problem that the calculated result fluctuates depending on the contents of the teacher data and the software used, and there is no reproducibility when reconstructing a neural circuit.
[0029]
In addition, in the conventional method using a neural network or the like, the operation process is treated as a black box, and the obtained result is not clear, that is, there is a problem that the validity of the calculation process of the result cannot be verified. .
[0030]
In addition, in the conventional clustering method, since the differences between samples and genes are expressed as relative ones within a given data group by grouping, the evaluation result is unstable. Had a point.
[0031]
Further, in the conventional method of grouping genes by “unsupervised”, it was difficult to verify the results, and thus had a problem that trial and error had to be repeated.
[0032]
In addition, the conventional method does not define a procedure for mechanically verifying which items are effective for discrimination from gene expression data or sample information, so that it is not possible to efficiently select an optimum item. Had a point.
[0033]
As described above, at the present time, the conventional system and the like have a number of problems, and as a result, both the user and the administrator are inconvenient and the utilization efficiency is low.
[0034]
Here, as a technique used for prediction and discrimination of various phenomena, there is an “MT (Maharanobis-Taguchi) system method” which is a technique for measuring all phenomena using Mahalanobis distance proposed by Dr. Genichi Taguchi (for example, Genichi Taguchi, "Technological Development in MT System" (Japan Standards Association, 2002)).
[0035]
The principle of the MT system method is as follows: (1) For a phenomenon to be measured, a data group with a small variation in data and results is further selected from a data group with a known result, and a database is created based on this data group. (2) This database is constructed by analyzing unknown phenomena as Mahalanobis distances as a "measure".
[0036]
Here, as described above, the MT system method is used for data analysis of each device for device monitoring and preventive maintenance (for example, see Patent Literature 1), and prediction of a health examination by the MT system method is performed. Although a method (for example, see Non-patent Document 13) is known, there is no report on applying the MT system method to gene data analysis in DNA microarray analysis.
[0037]
The present invention has been made in view of the above-mentioned problems, and a DNA microarray data analysis method, a DNA microarray data analysis apparatus, and a program capable of analyzing a large amount of gene data at high speed and with high accuracy by applying the MT system method. And a recording medium.
[Means for Solving the Problems]
[0038]
In order to achieve such an object, the DNA microarray data analysis method according to claim 1 is characterized by analyzing measurement data of a DNA microarray using a Mahalanobis-Taguchi system.
[0039]
According to this method, since the measurement data of the DNA microarray is analyzed using the Mahalanobis Taguchi system, a large amount of genetic data can be analyzed at high speed and with high accuracy by applying the MT system method.
[0040]
That is, in the conventional DNA microarray data analysis method using a neural network that requires teacher data, the features of all the teacher data (data for which the determination result is clear, such as the sensitivity of a clinically determined drug) are quantified. However, in the MT system method, the analysis can be started with reference to the vague contents such as whether or not the drug has an effect and whether or not the drug belongs to a specific group. Teacher data) Preparation becomes very simple.
[0041]
In the conventional DNA microarray data analysis method using neural network learning, multiple regression analysis coefficient calculation, and the like, the same operation is repeated a plurality of times to form neural circuits and parameters in an optimal state. In the method, a database (reference data) can be generated by one calculation and can be processed at a high speed, so that the operation time can be extremely shortened.
[0042]
Further, in the conventional DNA microarray data analysis method using a neural network or the like, the calculation process is treated as a black box, and the calculation process of the obtained result is not clear (that is, the validity of the calculation process cannot be verified). Although several calculation methods have been proposed in the MT system method, they are all open calculation formulas, and it is clear that the calculation formula focuses on the correlation of all reference data. Can be guaranteed.
[0043]
In the conventional DNA microarray data analysis method using a neural network or the like, the calculated result fluctuates depending on the contents of the teacher data and the software used, and there is no reproducibility when reconstructing a neural circuit. In the method, the calculation is performed using a published calculation formula, so that the same result can be always obtained by using the same data, and the reproducibility can be extremely improved.
[0044]
Further, in a DNA microarray data analysis method using a general clustering technique, a difference between a specimen and a gene is expressed as a relative one within a group by dividing the data into groups within a given data group. In the MT system method, once a unit space database is generated, it can be expressed as an absolute value by expressing it as a distance from the origin of this reference, and other than the analyzed data group, the Mahalanobis distance from the unit space database can be expressed. Can be evaluated simply by calculating the evaluation result, so that the evaluation result becomes stable every time.
[0045]
Furthermore, in the MT system method, a procedure for mechanically verifying which items (gene expression data and specimen information) are effective for discrimination is defined (item selection). The most appropriate items can be selected efficiently without being bound by biased opinions.
[0046]
Further, the DNA microarray data analysis method according to claim 2 is characterized in that genes to be analyzed are classified from the measurement data of the DNA microarray using a Mahalanobis-Taguchi system.
[0047]
According to this method, the gene to be analyzed is classified from the measurement data of the DNA microarray using the Mahalanobis-Taguchi system, so that a large amount of gene data can be classified at high speed and with high accuracy by applying the MT system method. Become like
[0048]
In addition, the DNA microarray data analysis method according to claim 3 selects a gene to be analyzed from the measurement data of the sample DNA microarray using a Mahalanobis Taguchi system, and uses the Mahalanobis Taguchi system to select the selected gene data. Classifying a clinical sample based on
[0049]
According to this method, a gene to be analyzed is selected from the measurement data of the DNA microarray of the sample using the Mahalanobis-Taguchi system, and the clinical sample is classified based on the selected gene data using the Mahalanobis-Taguchi system. By applying the MT system method, a large amount of microarray measurement data can be classified at high speed and with high accuracy.
[0050]
In addition, the DNA microarray data analysis method according to claim 4 includes a data group selection step of selecting a data group by collecting reference data according to a discrimination item to be analyzed from measurement data of the DNA microarray, An analysis data item determination step for determining an analysis data item which is an item to be input data when discriminated by the MT system method and which is an item analyzed in the MT system method, and which is selected in the data group selection step. A unit space database creation step of creating a unit space database indicating a Mahalanobis space of the reference data included in the reference data group, a Mahalanobis distance of the reference data selected in the data group selection step, and the reference data Mahalanobis distance by collecting comparison data that does not clearly belong to Based on the Mahalanobis distance calculation step of calculating the Mahalanobis distance between the reference data and the comparison data calculated in the Mahalanobis distance calculation step, a threshold that is a boundary between the reference data group and the comparison data group A threshold setting step of setting one or more, and a Mahalanobis distance of unknown data is calculated by using the unit space database and the selected item, and discrimination analysis of unknown data is performed by determining the Mahalanobis distance by the threshold. Is performed.
[0051]
According to this method, a reference data group is selected by collecting reference data corresponding to a discrimination item to be analyzed from measurement data of the DNA microarray, and an item serving as input data when discrimination is performed by the MT system method. And determining an analysis data item that is an item analyzed in the MT system method, creating a unit space database indicating a Mahalanobis space of the reference data included in the selected reference data group, and selecting the selected reference data and Calculate the Mahalanobis distance of comparison data that does not clearly belong to the reference data, and set one or more thresholds that serve as boundaries between the reference data group and the comparison data group based on the calculated Mahalanobis distance between the reference data and the comparison data The Mahalanobis distance of unknown data is calculated using the unit space database and the selected item, Since classify unknown data by determining the value, by applying the MT system method, it is possible to analyze quickly and accurately a large amount of genetic data.
[0052]
That is, in the MT system method, the analysis can be started with the vague contents of whether or not the medicine has an effect or whether or not to select a specific group as reference data. Becomes very simple.
[0053]
Further, in the MT system method, a database (reference data) can be generated by one calculation and can be processed at a high speed, so that the operation time can be extremely shortened.
[0054]
Although several calculation methods have been proposed in the MT system method, they are all open calculation formulas, and it is clear that the calculation formula focuses on the correlation of all reference data. Clarity can be ensured.
[0055]
Further, in the MT system method, since the calculation is performed using a publicly available calculation formula, the same result can be always obtained by using the same data, and the reproducibility can be extremely improved.
[0056]
Also, in the MT system method, once a unit space database is generated, it can be expressed as an absolute value by expressing it as a distance from the origin of this reference, and other than the analyzed data group, the Mahalanobis from the unit space database can be used. Since the evaluation can be performed only by calculating the distance, the evaluation result becomes stable every time. Also, in the MT system method, when a causal relationship is confirmed between the phenomenon indicated by the target and the Mahalanobis distance, by setting a plurality of thresholds, it is possible to classify a plurality of states simply by calculating the Mahalanobis distance. I do.
Further, in the MT system method, when a certain proportional relationship is established between the target state and the Mahalanobis distance, the target state itself can be expressed by the Mahalanobis distance.
[0057]
Furthermore, in the MT system method, a procedure for mechanically verifying which items (gene expression data and specimen information) are effective for discrimination is defined (item selection). The most appropriate items can be selected efficiently without being bound by biased opinions.
[0058]
In the DNA microarray data analysis method according to a fifth aspect, in the DNA microarray data analysis method according to the fourth aspect, data that belongs to the reference data besides the reference data group is separately prepared, A verification data group in which data that does not belong to the data is mixed is created, the Mahalanobis distance is calculated for the verification data group, and the reference data group and the other data are calculated using the threshold set in the threshold setting step. And a discrimination accuracy verification step of verifying whether or not a group can be correctly discriminated.
[0059]
This more specifically shows an example of the discrimination accuracy verification step. According to this method, data that should belong to the reference data besides the reference data group is separately prepared, a verification data group in which this data is mixed with data that does not belong to the reference data is created, and the Mahalanobis distance is calculated for the verification data group Since it is verified whether the reference data group can be correctly distinguished from the other groups using the set threshold value, the determination operation can be performed more efficiently.
[0060]
Further, in the DNA microarray data analysis method according to claim 6, in the DNA microarray data analysis method according to claim 5, when it is determined that sufficient accuracy cannot be obtained in the verification step, the data group selection is performed. The method further includes a review step of reviewing the reference data group selected in the step and / or the analysis data item determined in the analysis data item determination step.
[0061]
This shows one example of the review step more specifically. According to this method, when it is determined that sufficient accuracy cannot be obtained, the selected reference data group and / or the determined analysis data item are re-examined, so that the determination operation can be performed more accurately and efficiently. Will be able to do it.
[0062]
The DNA microarray data analysis method according to claim 7 is the DNA microarray data analysis method according to claim 6, wherein the reviewing step includes the standard deviation of each item in the reference data group and the frequency of data appearance. The method further includes a reference data reviewing step of verifying the degree of variation of the reference data using the reference data and using the reference data having a small degree of variation.
[0063]
This shows one example of the reference data review step more specifically. According to this method, the degree of variation of the reference data is verified using the standard deviation and the frequency of data appearance of each item in the reference data group, and the reference data having the small degree of variation is used. By reconsidering, the discrimination work can be performed more accurately and efficiently.
[0064]
In the DNA microarray data analysis method according to claim 8, in the DNA microarray data analysis method according to claim 6 or 7, the re-examination step includes determining which spot measurement data of the DNA microarray is to be used. The method further includes an analysis data item reexamination step of verifying a plurality of combinations using an orthogonal table, evaluating the S / N ratio, and then using an analysis data item having a higher S / N ratio as the analysis data item. It is characterized by including.
[0065]
This more specifically shows an example of the analysis data item review step. According to this method, which combination of spots on the DNA microarray is used to verify the measurement data is verified using a plurality of combinations using an orthogonal table. Since the analysis data item is used, the discrimination work can be performed more accurately and efficiently by reviewing the analysis data item.
[0066]
The present invention also relates to a DNA microarray data analyzer, wherein the DNA microarray data analyzer according to claim 9 converts reference data according to a discrimination item to be analyzed from among the measurement data of the DNA microarray. Data group selecting means for selecting a data group by collecting, and analysis data for determining an analysis data item which is an item serving as input data when discriminated by the MT system method and which is an item analyzed by the MT system method. Item determination means, unit space database creation means for creating a unit space database indicating a Mahalanobis space of the reference data included in the reference data group selected by the data group selection means, and the data group selection means The Mahalanobis distance of the selected reference data and the reference data Mahalanobis distance calculating means for collecting Mahalanobis distance by collecting comparison data that does not belong to, based on the Mahalanobis distance between the reference data and the comparison data calculated by the Mahalanobis distance calculating means, the reference data group and Using a threshold setting means for setting one or a plurality of thresholds serving as boundaries of the comparison data group, and using the unit space database and the selected item, calculate a Mahalanobis distance of unknown data. Classification means for performing discriminant analysis of unknown data by making a determination is provided.
[0067]
According to this apparatus, a reference data group is selected by collecting reference data corresponding to a discrimination item to be analyzed from measurement data of a DNA microarray, and an item serving as input data when discrimination is performed by the MT system method. And determining an analysis data item that is an item analyzed in the MT system method, creating a unit space database indicating a Mahalanobis space of the reference data included in the selected reference data group, and selecting the selected reference data and Calculate the Mahalanobis distance of comparison data that does not clearly belong to the reference data, and set one or more thresholds that serve as boundaries between the reference data group and the comparison data group based on the calculated Mahalanobis distance between the reference data and the comparison data The Mahalanobis distance of unknown data is calculated using the unit space database and the selected item, Since classify unknown data by determining the value, by applying the MT system method, it is possible to analyze quickly and accurately a large amount of genetic data.
[0068]
That is, in the MT system method, the analysis can be started with the vague contents of whether or not the medicine has an effect or whether or not to select a specific group as reference data. Becomes very simple.
[0069]
Further, in the MT system method, a database (reference data) can be generated by one calculation and can be processed at a high speed, so that the operation time can be extremely shortened.
[0070]
Although several calculation methods have been proposed in the MT system method, they are all open calculation formulas, and it is clear that the calculation formula focuses on the correlation of all reference data. Clarity can be ensured.
[0071]
Further, in the MT system method, since the calculation is performed using a publicly available calculation formula, the same result can be always obtained by using the same data, and the reproducibility can be extremely improved.
[0072]
Also, in the MT system method, once a unit space database is generated, it can be expressed as an absolute value by expressing it as a distance from the origin of this reference, and other than the analyzed data group, the Mahalanobis from the unit space database can be used. Since the evaluation can be performed only by calculating the distance, the evaluation result becomes stable every time. Also, in the MT system method, when a causal relationship is confirmed between the phenomenon indicated by the target and the Mahalanobis distance, by setting a plurality of thresholds, it is possible to classify a plurality of states simply by calculating the Mahalanobis distance. I do.
Further, in the MT system method, when a certain proportional relationship is established between the target state and the Mahalanobis distance, the target state itself can be expressed by the Mahalanobis distance.
[0073]
Furthermore, in the MT system method, a procedure for mechanically verifying which items (gene expression data and specimen information) are effective for discrimination is defined (item selection). The most appropriate items can be selected efficiently without being bound by biased opinions.
[0074]
The DNA microarray data analyzer according to claim 10 is the DNA microarray data analyzer according to claim 9, wherein data that belongs to the reference data is separately prepared in addition to the reference data group. A verification data group in which data that does not belong to the data is mixed is created, the Mahalanobis distance is calculated for the verification data group, and the reference data group and the other data are calculated using the threshold set by the threshold setting unit. A determination accuracy verification unit configured to verify whether the group can be correctly determined.
[0075]
This shows an example of the discrimination accuracy verification means more specifically. According to this device, data that should belong to the reference data besides the reference data group is separately prepared, a verification data group in which this data and data that does not belong to the reference data are mixed is created, and the Mahalanobis distance is calculated for the verification data group. Since it is verified whether the reference data group can be correctly distinguished from the other groups using the set threshold value, the determination operation can be performed more efficiently.
[0076]
Further, in the DNA microarray data analyzer according to claim 11, when the DNA microarray data analyzer according to claim 10 determines that sufficient accuracy cannot be obtained by the verification means, the data group selection is performed. A reexamination means for reexamining the reference data group selected by the means and / or the analysis data item determined by the analysis data item determination means is further provided.
[0077]
This more specifically shows an example of the reviewing means. According to this device, when it is determined that sufficient accuracy cannot be obtained, the selected reference data group and / or the determined analysis data item are reexamined, so that the determination operation can be performed with higher accuracy and efficiency. Will be able to do it.
[0078]
The DNA microarray data analyzer according to claim 12 is the DNA microarray data analyzer according to claim 11, wherein the reviewing means includes a standard deviation of each item in the reference data group and a data appearance frequency. And a reference data reviewing means for verifying the degree of variation of the reference data using the reference data having a small degree of variation.
[0079]
This more specifically shows one example of the reference data reviewing means. According to this device, the degree of variation of the reference data is verified using the standard deviation and the frequency of data appearance of each item in the reference data group, and the reference data having the small degree of variation is used. By reconsidering, the discrimination work can be performed more accurately and efficiently.
[0080]
The DNA microarray data analyzer according to claim 13 is the DNA microarray data analyzer according to claim 11 or 12, wherein the reviewing means determines which spot of the DNA microarray uses the measurement data. After analyzing in a plurality of combinations using an orthogonal table and evaluating with an SN ratio, an analysis data item reexamination means for using an item having a higher SN ratio when used as the analysis data item as the analysis data item is further provided. It is characterized by having.
[0081]
This more specifically shows an example of the analysis data item reviewing means. According to this apparatus, which combination of spots on the DNA microarray is to be used for measurement data is verified using a plurality of combinations using an orthogonal table. Since the analysis data item is used, the discrimination work can be performed more accurately and efficiently by reviewing the analysis data item.
[0082]
The present invention also relates to a program for causing a computer to execute a method for analyzing DNA microarray data, wherein the program according to claim 14 is a program for measuring, based on measurement data of a DNA microarray, a standard according to a discrimination item to be analyzed. A data group selection step of selecting a data group by collecting data; and an analysis data item which is an item serving as input data when discriminated by the MT system method and which is an item analyzed by the MT system method. An analysis data item determination step, a unit space database creation step of creating a unit space database indicating a Mahalanobis space of the reference data included in the reference data group selected in the data group selection step, and the data group selection step Maha of the above reference data selected in A Novice distance, a Mahalanobis distance calculation step of calculating a Mahalanobis distance by collecting comparison data that does not clearly belong to the reference data, and the Mahalanobis distance between the reference data and the comparison data calculated in the Mahalanobis distance calculation step A threshold setting step of setting one or a plurality of thresholds as boundaries between the reference data group and the comparison data group based on the above, and using the unit space database and the selected item, the Mahalanobis distance of unknown data , And making the computer execute a DNA microarray data analysis method including a classification step of performing discrimination analysis of unknown data by judging this by the threshold value.
[0083]
According to this program, a reference data group is selected by collecting reference data according to a discrimination item to be analyzed from measurement data of a DNA microarray, and an item serving as input data when discrimination is performed by the MT system method. And determining an analysis data item that is an item analyzed in the MT system method, creating a unit space database indicating a Mahalanobis space of the reference data included in the selected reference data group, and selecting the selected reference data and Calculate the Mahalanobis distance of comparison data that does not clearly belong to the reference data, and set one or more thresholds that serve as boundaries between the reference data group and the comparison data group based on the calculated Mahalanobis distance between the reference data and the comparison data Then, the Mahalanobis distance of unknown data is calculated using the unit space database and the selected items. Since classify unknown data by determining by the threshold, by application of an MT system method, it is possible to analyze quickly and accurately a large amount of genetic data.
[0084]
That is, in the MT system method, the analysis can be started with the vague contents of whether or not the medicine has an effect or whether or not to select a specific group as reference data. Becomes very simple.
[0085]
Further, in the MT system method, a database (reference data) can be generated by one calculation and can be processed at a high speed, so that the operation time can be extremely shortened.
[0086]
Although several calculation methods have been proposed in the MT system method, they are all open calculation formulas, and it is clear that the calculation formula focuses on the correlation of all reference data. Clarity can be ensured.
[0087]
Further, in the MT system method, since the calculation is performed using a publicly available calculation formula, the same result can be always obtained by using the same data, and the reproducibility can be extremely improved.
[0088]
Also, in the MT system method, once a unit space database is generated, it can be expressed as an absolute value by expressing it as a distance from the origin of this reference, and other than the analyzed data group, the Mahalanobis from the unit space database can be used. Since the evaluation can be performed only by calculating the distance, the evaluation result becomes stable every time. Also, in the MT system method, when a causal relationship is confirmed between the phenomenon indicated by the target and the Mahalanobis distance, by setting a plurality of thresholds, it is possible to classify a plurality of states simply by calculating the Mahalanobis distance. I do.
Further, in the MT system method, when a certain proportional relationship is established between the target state and the Mahalanobis distance, the target state itself can be expressed by the Mahalanobis distance.
[0089]
Furthermore, in the MT system method, a procedure for mechanically verifying which items (gene expression data and specimen information) are effective for discrimination is defined (item selection). The most appropriate items can be selected efficiently without being bound by biased opinions.
[0090]
A program according to a fifteenth aspect of the present invention is the program according to the fourteenth aspect, wherein data which belongs to the reference data is separately prepared in addition to the reference data group, and data which does not belong to the reference data is mixed. A verification data group is created, the Mahalanobis distance is calculated for the verification data group, and it is verified whether the reference data group can be correctly distinguished from the other groups using the threshold set in the threshold setting step. And a verification accuracy verification step.
[0091]
This more specifically shows an example of the discrimination accuracy verification step. According to this program, data that should belong to the reference data besides the reference data group is separately prepared, a verification data group is created by mixing this with data that does not belong to the reference data, and the Mahalanobis distance is calculated for the verification data group Since it is verified whether the reference data group can be correctly distinguished from the other groups using the set threshold value, the determination operation can be performed more efficiently.
[0092]
The program according to claim 16 is the program according to claim 15, wherein when it is determined that sufficient accuracy is not obtained in the verification step, the reference data selected in the data group selection step is selected. The method further includes a review step of reviewing the group and / or the analysis data item determined in the analysis data item determination step.
[0093]
This shows one example of the review step more specifically. According to this program, when it is determined that sufficient accuracy cannot be obtained, the selected reference data group and / or the determined analysis data items are re-examined, so that the determination operation can be performed with higher accuracy and efficiency. Will be able to do it.
[0094]
The program according to claim 17 is the program according to claim 16, wherein the reviewing step is performed by using a standard deviation of each item in the reference data group, a data appearance frequency, and the like to determine how the reference data varies. And further including a reference data reviewing step of using the reference data having a small degree of variation.
[0095]
This shows one example of the reference data review step more specifically. According to this program, the degree of variation of the reference data is verified using the standard deviation and the frequency of appearance of each item in the reference data group, and the reference data having the small degree of variation is used. By reconsidering, the discrimination work can be performed more accurately and efficiently.
[0096]
The program according to claim 18 is the program according to claim 16 or 17, wherein the re-examination step includes a plurality of combinations using the orthogonal table as to which measurement data of which spot on the DNA microarray is to be used. The method further includes an analysis data item reexamination step of using, as an analysis data item, an item having a high SN ratio when the analysis data item is used as the analysis data item after the verification with the SN ratio.
[0097]
This more specifically shows an example of the analysis data item review step. According to this program, which combination of spots on the DNA microarray is used to verify the data to be measured is examined in multiple combinations using an orthogonal table. Since the analysis data item is used, the discrimination work can be performed more accurately and efficiently by reviewing the analysis data item.
[0098]
The present invention also relates to a recording medium, wherein a recording medium according to claim 19 records the program according to any one of claims 14 to 18.
[0099]
According to this recording medium, a program recorded on the recording medium is read by a computer and executed, thereby realizing the program according to any one of claims 14 to 18 using a computer. And the same effect as each of these methods can be obtained.
[0100]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of a DNA microarray data analysis method, a DNA microarray data analysis device, a program, and a recording medium according to the present invention will be described in detail with reference to the drawings. It should be noted that the present invention is not limited by the embodiment.
In particular, in the following embodiment, an example in which the present invention is applied to prediction of a DNA microarray experiment result regarding interferon (INF) sensitivity will be described. However, the present invention is not limited to this case. The same can be applied.
[0101]
[Summary of the present invention]
Hereinafter, the outline of the present invention will be described, and then the configuration, processing, and the like of the present invention will be described in detail. FIG. 1 is a principle configuration diagram showing the basic principle of the present invention.
The present invention generally has the following basic features. The MT system method is a technology for measuring all phenomena using the Mahalanobis distance proposed by Dr. Genichi Taguchi. The principle of the MT system is that, for a phenomenon to be measured, a data group (reference data group) with less variation in data and results is selected from among data groups whose results are known, and based on this, the correlation of all input data is selected. Create the database shown. Then, by using this database as a "measure" and analyzing unknown phenomena as Mahalanobis distances, it is a technique for realizing prediction and discrimination of various phenomena.
[0102]
When the MT system method is applied to the analysis of the measurement data of the DNA microarray, the test result data of the DNA microarray + specimen data and the like are input data, and the predicted state of the specimen or a change predicted in the future is the analysis result. . Then, in the process of grouping the samples, the analysis target genes are narrowed down (searching for effective genes) while checking the accuracy from the beginning.
[0103]
Here, two types of conventional DNA microarray analysis techniques are generally used. One is a technology related to quantification (quantification) of analysis results (for example, a technology based on neural network, multiple regression analysis, fuzzy theory, etc.), and the other is grouping based on quantified data. Technology.
[0104]
Conventionally, in DNA microarray analysis, data analysis processing has been performed using these two techniques in two stages. For example, the quantification (similarization) of the similarity between genes performed in Non-Patent Document 1 corresponds to the quantification (numericalization) here, and is used as an input value for subsequent clustering (grouping). Therefore, a large amount of calculation time is required for analyzing a large amount of data covering, for example, hundreds to tens of thousands of items, and the analysis accuracy is not very high. Further, although the narrowing down of genes in Non-Patent Document 2 is not a two-step process as described above, grouping by a neural network is performed an enormous number of times, so that it is expected that ordinary computer calculations will take several days.
[0105]
Here, the relationship between the quantification and grouping described above and the category determination and assignment to the category described in the “prior art” is described. In each of the category determination and category assignment, the quantification and grouping are performed. It can be said that there is. However, as in the neural network of Non-Patent Document 2, not all methods are necessarily divided into stages, and there may be a combination of both in one process.
[0106]
In contrast, in the MT system, a database is created and used for subsequent calculations. Therefore, by using the MT system for DNA microarray analysis, genes can be narrowed down while checking the accuracy in the process of grouping samples from the beginning. Therefore, it is possible to perform the above two arithmetic processings at a high accuracy and at a high speed. As a result, it is possible to efficiently perform the classification of the specimen based on the gene expression state and the role analysis of the gene itself.
[0107]
In other words, by using the MT system method for DNA microarray analysis, it is possible to represent the position of each specimen by the Mahalanobis distance from the reference data group. And the efficiency of analysis can be improved. As a result, it is possible to efficiently search for a gene or specimen information that is considered effective.
[0108]
Further, by making the correlation of all the input data into a database, it becomes possible to catch a phenomenon that cannot be detected with only single data.
[0109]
Thus, the present inventors have found that the above-described MT system method is applied to DNA microarray analysis. Therefore, the present inventor performed analysis by the MT system method using commercially available software having a data analysis function by the MT system method (for example, “PRAT” (product name) of a probe (company name)). Some commercially available products are available as software for the MT system method, and any of them can be used.
[0110]
However, there are problems to be overcome in order to actually perform DNA microarray analysis using commercially available software. That is, in the MT system method, it is established as common general knowledge for those skilled in the art that the number of reference data must be collected more than the number of items. The number of reference data corresponds to the number of individual samples in the present invention, and the number of items corresponds to the number of DNA microarray-equipped genes in the present invention plus the number of clinical data such as blood tests. It is almost impossible to adapt to such conditions in the analysis of.
[0111]
Further, when a DNA microarray analysis was actually performed using commercially available software, it was found that many genes with extremely low expression levels were present, and that the CV (coefficient of variation) between microarrays was unstable. It was difficult to apply the MT system method because of the problems inherent in the analysis.
[0112]
Therefore, the present inventor has found that for the former problem, by selecting items with higher characteristics in advance, it is possible to create an effective unit space even from a reference data number extremely smaller than the number of items, Solved this. Regarding the latter problem, a combination of items in an orthogonal table was tried, and the calculation results were compared based on the S / N ratio to select an effective item.
[0113]
Furthermore, in the MT system method, at the time of creating a unit space, it is necessary to prepare two types of a reference data group as a source of the unit space and an abnormal data group that can be clearly identified as "abnormal". (See "Technological Development in MT Systems," published by the Japan Standards Association, pages 428-429, "Setting Normal and Abnormal Data for Measurers.") In clinical practice, the data is ambiguous and it is clear which data group is included. Sample is not included, and it is often difficult to judge. Therefore, the present inventor has first devised a unit space using only the reference data and then devised with the comparison data of the evaluation, thereby devising the analysis while allowing the existence of the sample for which the determination is suspended.
[0114]
For example, when applied to the prediction of DNA microarray experiment results for interferon (INF) sensitivity, using the above-described device, the results of 18 samples out of 88 samples were selected as reference data, and the remaining 70 samples were verified. As a result, it was found that the INF sensitivity of the patient can be expected with high frequency (80 to 90%). The details of these ideas and other improvements will be described later.
[0115]
As described above, simply applying the MT system method cannot be realized as a practically effective one, and based on the results of the inventors' ingenuity, the application of the MT system method in DNA microarray analysis has been considered. Became practically effective.
[0116]
[System configuration]
First, the configuration of the present system will be described. FIG. 2 is a block diagram showing an example of the configuration of the present system to which the present invention is applied, and conceptually shows only those parts of the configuration related to the present invention. The present system roughly connects a DNA microarray data analyzer 100 and an external system 200 that provides an external database or homology search and other external programs related to sequence information and the like via a network 300 so as to be communicable. It is configured.
[0117]
In FIG. 2, a network 300 has a function of interconnecting the DNA microarray data analyzer 100 and the external system 200, and is, for example, the Internet.
[0118]
In FIG. 2, an external system 200 is mutually connected to the DNA microarray data analyzer 100 via a network 300, and executes an external database for sequence information and the like, and an external program such as a homology search or a motif search for a user. Has the function of providing a website.
[0119]
Here, the external system 200 may be configured as a WEB server, an ASP server, or the like, and its hardware configuration may be configured by an information processing device such as a generally-available workstation, a personal computer, and its accompanying devices. Good. Each function of the external system 200 is realized by a CPU, a disk device, a memory device, an input device, an output device, a communication control device, and the like in a hardware configuration of the external system 200, a program for controlling them, and the like.
[0120]
In FIG. 2, a DNA microarray data analyzer 100 schematically includes a control unit 102 such as a CPU that integrally controls the entire DNA microarray data analyzer 100, and a communication device such as a router connected to a communication line or the like (see FIG. 2). (Not shown), an input / output control interface unit 108 connected to the input device 112 and the output device 114, and a storage unit 106 for storing various databases and tables. These units are communicably connected via an arbitrary communication path. Further, the DNA microarray data analyzer 100 is communicably connected to the network 300 via a communication device such as a router and a wired or wireless communication line such as a dedicated line.
[0121]
Various databases and tables (DNA microarray measurement data file 106a to verification data group file 106g) stored in the storage unit 106 are storage means such as a fixed disk device, and various programs, tables and files used for various processes. Stores database and web page files.
[0122]
Among the constituent elements of the storage unit 106, the DNA microarray measurement data file 106a is a file storing DNA microarray measurement data (for example, test result data such as expression data of each gene and specimen data). FIG. 5 is a diagram showing an example of the user information stored in the DNA microarray measurement data file 106a.
[0123]
As shown in FIG. 5, the information stored in the DNA microarray measurement data file 106a includes a sample ID for uniquely identifying each sample, test result data such as expression data of each gene, and sample data. It is configured in association with. Here, the sample data is configured to include items of medical information (for example, liver function data by a blood test).
[0124]
The reference data group file 106b is a reference data group storage unit that stores information on the reference data group and the like.
[0125]
The comparison data group file 106c is a comparison data group storage unit that stores information on the comparison data group and the like.
[0126]
The analysis data item 106d is an analysis data item storage unit that stores information on the analysis data item and the like.
[0127]
Further, the unit space database 106e obtains a correlation coefficient, a standard deviation, and the like from each reference data included in the selected reference data group, and stores information on a database indicating the unit space (Maharanobis space) and the like. Storage means.
[0128]
The Mahalanobis distance file 106f is a Mahalanobis distance data storage unit that stores information on the Mahalanobis distance for each of the reference data and the comparison data.
[0129]
The verification data group file 106g is a verification data group storage unit that stores information on the verification data group and the like.
[0130]
In FIG. 2, the communication control interface unit 104 controls communication between the DNA microarray data analysis device 100 and the network 300 (or a communication device such as a router). That is, the communication control interface unit 104 has a function of communicating data with another terminal via a communication line.
[0131]
2, the input / output control interface unit 108 controls the input device 112 and the output device 114. Here, as the output device 114, in addition to a monitor (including a home television), a speaker can be used (in the following, the output device 114 may be described as a monitor). As the input device 112, a keyboard, a mouse, a microphone, and the like can be used. The monitor also realizes a pointing device function in cooperation with the mouse.
[0132]
2, the control unit 102 has a control program such as an OS (Operating System), a program defining various processing procedures and the like, and an internal memory for storing required data. And information processing for executing various processes. The control unit 102 is functionally conceptually composed of an analysis purpose determination unit 102a, a data group selection unit 102b, an analysis data item determination unit 102c, a unit space database creation unit 102d, a Mahalanobis distance calculation unit 102e, a threshold setting unit 102f, and a judgment accuracy verification. It comprises a unit 102g, a review unit 102h, and a determination unit 102i.
[0133]
The analysis purpose determination unit 102a is an analysis purpose determination unit that determines an item to be determined based on the data of the DNA microarray when analyzing the measurement data of the DNA microarray by the MT system method.
[0134]
Further, the data group selection unit 102b selects a reference data group by collecting reference data according to the discrimination item to be analyzed from the measurement data of the DNA microarray, and further selects a comparison data group that does not clearly belong to the above reference data. Data group selecting means for selecting a comparison data group by collecting data.
[0135]
The analysis data item determination unit 102c is an analysis data item determination unit that determines an analysis data item that is input data when the determination is performed by the MT system method and is an item that is analyzed by the MT system method. .
[0136]
The unit space database creation unit 102d is a unit space database creation unit that creates a unit space database indicating a Mahalanobis space of the reference data included in the reference data group selected by the data group selection unit 102b.
[0137]
The Mahalanobis distance calculation unit 102e is a Mahalanobis distance calculation unit that calculates a Mahalanobis distance between the reference data selected by the data group selection unit 102b and the comparison data.
[0138]
The threshold setting unit 102f is a threshold setting unit that sets a threshold that is a boundary between the reference data group and the comparison data group based on the Mahalanobis distance between the reference data and the comparison data calculated by the Mahalanobis distance calculation unit 102e. is there.
[0139]
In addition, the judgment accuracy verification unit 102g separately prepares data that should belong to the reference data in addition to the reference data group, creates a verification data group in which this data and data that does not belong to the reference data are mixed, and creates a Mahalanobis for the verification data group. This is a discrimination accuracy verification unit that calculates the distance and verifies whether the reference data group and the other groups can be correctly discriminated using the threshold set by the threshold setting unit 102f.
[0140]
When the judgment accuracy verification unit 102g determines that sufficient accuracy cannot be obtained, the review unit 102h determines whether the reference data group selected by the data group selection unit 102b and / or the analysis data item determination unit This is a reviewing means for reviewing the analysis data item determined in 102c. Here, FIG. 6 is a block diagram illustrating an example of the configuration of the review unit 102h. As illustrated in FIG. 6, the review unit 102h includes a reference data review unit 102j and an analysis data item review unit 102k.
[0141]
Here, the reference data reviewing unit 102j verifies the degree of variation of the reference data using the standard deviation and the data appearance frequency of each item in the reference data group, and uses the reference data with the small degree of variation. It is a means of reviewing reference data.
[0142]
The analysis data item reviewing unit 102k verifies which spot of the DNA microarray to use for the measurement data in a plurality of combinations using an orthogonal table, evaluates the S / N ratio, and uses the S / N ratio as an analysis data item. This is an analysis data item reexamination means for using items having a higher value as analysis data items.
In addition, the above-mentioned function includes a function of examining an item using another statistical method or using the above method in combination.
[0143]
Returning to FIG. 2 again, the determination unit 102i is a determination unit that determines the unit space database and the threshold when it is determined that sufficient accuracy can be ensured as a result of the verification by the determination accuracy verification unit 102g.
[0144]
The details of the processing performed by these units will be described later.
[0145]
[System processing]
Next, an example of the processing of the present system configured as described above according to the present embodiment will be described in detail below with reference to FIG.
[0146]
[DNA microarray analysis processing by MT system]
First, details of the DNA microarray analysis processing by the MT system will be described with reference to FIG. FIG. 3 is a flowchart illustrating an example of a DNA microarray analysis process by the MT system of the present system in the present embodiment.
[0147]
(1) Analysis purpose determination (purpose definition)
First, the DNA microarray data analyzer 100 determines an item (discrimination item) to be determined based on the DNA microarray data when the measurement data of the DNA microarray is analyzed by the MT system method by the processing of the analysis purpose determination unit 102a. The purpose of the analysis is clarified (step SA-1).
[0148]
For example, the discrimination item for determining whether interferon administration is effective or ineffective for a patient with hepatitis C includes the effectiveness of interferon administration.
[0149]
(2) Selection of reference data group + comparison data group (reference data, comparison data selection)
In the MT system, in order to use a reference group as a database, a group corresponding to a discrimination item to be analyzed must be collected and used as original data (reference data group) of the database. Further, in addition to the reference data group, it is necessary to collect a data group different from the reference for comparison and select this as a comparison data group. The analysis accuracy is determined based on whether these groups can be accurately determined.
[0150]
Therefore, the DNA microarray data analyzer 100 accesses the DNA microarray measurement data file 106a by the processing of the data group selection unit 102b, and, from the DNA microarray measurement data, sets the reference data according to the discrimination item to be analyzed. To select a reference data group, and a comparison data group is selected by collecting comparison data that does not clearly belong to the reference data (step SA-2).
[0151]
Here, it is desirable that the population selected as the reference data be data with as little variation as possible, and it is desirable to select a population that can have a large number as a population.
[0152]
For example, when analyzing the efficacy of interferon administration, if the administration results are known from the data held, but there are many specimens for which administration is effective, select the effective specimen as reference data, and Data group. The reverse is true if there are many invalid samples. For the comparison data group, a group that does not clearly belong to the reference data group is selected. That is, in this example, the “reference data” is an individual patient sample, and the “reference data group” is a group of patients (each group with and without IFN sensitivity).
[0153]
Then, the data group selection unit 102b stores the selected reference data group in a predetermined storage area of the reference data group file 106b, and stores the selected comparison data group in a predetermined storage area of the comparison data group file 106c. .
[0154]
(3) Analysis data item determination (input item determination)
Then, the DNA microarray data analysis device 100 accesses the DNA microarray measurement data file 106a to the comparison data group file 106c by the processing of the analysis data item determination unit 102c and becomes “input data” when the MT system makes a determination. An item (also referred to as a feature or a parameter), that is, an “analysis data item” which is an item analyzed in the MT system is determined (step SA-3).
[0155]
At this time, the items that result in the analysis must not be included in the analysis data items. For example, when it is desired to analyze the effectiveness of interferon administration, the measurement results (each gene data) of the DNA microarray and the items such as the age and gender of the sample are input data (analysis data items).
[0156]
Then, the comparison data group file 106c stores the determined analysis data item in a predetermined storage area of the analysis data item 106d.
[0157]
(4) Unit space database creation
Then, the DNA microarray data analyzer 100 accesses the reference data group file 106b and the like by the processing of the unit space database creating unit 102d, and obtains a correlation coefficient, a standard deviation, etc. from each reference data included in the selected reference data group. And a database showing these unit spaces (Maharanobis space) is created (step SA-4).
[0158]
In the MT system, there are several types of calculation methods, for example, a method using an inverse matrix, an orthogonal expansion by Schmidt, a method using a cofactor, and the like. In the present invention, any method may be used.
[0159]
As an example, a case where calculation is performed using an inverse matrix is shown below. Calculate the mean value and standard deviation value of all items from the reference data compiled in the group for which interferon administration was known to be effective, normalize each data, and calculate the correlation matrix and its inverse matrix from these. . Among them, the average value, the standard deviation value, and the inverse matrix form the unit space database. (See “Procedure 3” on page 5, “7.3.1 Mahalanobis Distance” on page 5 of the above-mentioned reference, “Technical Development in MT System”).
[0160]
Then, the unit space database creating unit 102d stores the created unit space database in a predetermined storage area of the unit space database 106e.
[0161]
(5) Calculation of Mahalanobis distance between reference data and comparison data
Then, the DNA microarray data analyzing apparatus 100 accesses the unit space database 106e and the like by the processing of the Mahalanobis distance calculation unit 102e, and obtains the reference data selected in step SA-2 for the unit space database created in step SA-4. Then, the Mahalanobis distance for each of the comparison data is calculated (step SA-5). That is, by this processing, the distance of each patient can be calculated. This distance is a measurement result by the MT system method. Here, there are several calculation methods for calculating the Mahalanobis distance, but the present invention may use any calculation method.
[0162]
As an example, a case where calculation is performed using an inverse matrix will be described below. The Mahalanobis distance is calculated from the unit space database created in (4) and each data value. This value indicates the effectiveness for interferon administration (see “Procedure 3” on page 5, “7.3.1 Mahalanobis distance” on page 5, “Technical development in MT system”, which is the above-mentioned reference). See.).
[0163]
Then, the Mahalanobis distance calculation unit 102e stores each calculated Mahalanobis distance in a predetermined storage area of the Mahalanobis distance file 106f.
[0164]
(6) Threshold setting
Then, the DNA microarray data analyzer 100 accesses the Mahalanobis distance file 106f and the like by the processing of the threshold setting unit 102f, and based on the Mahalanobis distances of the reference data calculated in step SA-5 and the comparison data, respectively. A value (threshold) serving as a boundary between the group and the comparison data group is set (step SA-6).
[0165]
Specifically, as shown in FIG. 4, a case where the distribution state is visually represented by a graph such as a histogram and verification is performed, and a case where the distribution state of the distance is verified by mathematical analysis is performed. Is mentioned. Here, FIG. 4 is a diagram showing an example of a case where a reference data group and a comparison data group are displayed in a distance histogram and a threshold value is set.
[0166]
Here, one or more thresholds may be set. In the MT system method, when a causal relationship is confirmed between the phenomenon indicated by the target and the Mahalanobis distance, by setting a plurality of thresholds, it becomes possible to classify a plurality of states simply by calculating the Mahalanobis distance. .
[0167]
As a specific example, when a proportional relationship is established between the Mahalanobis distance and the interferon administration effect, the Mahalanobis distance itself is used as the interferon administration effect prediction value, and nine thresholds are provided for dividing the prediction value into 10 groups. It may be used as an index for predicting the effect of administration of interferon in step evaluation.
[0168]
Further, when there are a plurality of groups other than the reference data group, the groups may be divided by the magnitude of the Mahalanobis distance to determine the plurality of groups. For example, in the right figure, the reference data group and the comparison data group intersect. Assuming that the reference data is a group in which the interferon administration is effective and the comparison data is an ineffective group, if it is desired to discriminate both with a higher probability, the Mahalanobis distance of this intersection may be set as a threshold.
[0169]
(7) Verification of discrimination accuracy using verification data group
Then, the DNA microarray data analyzer 100 accesses the DNA microarray measurement data file 106a or the like by the processing of the judgment accuracy verification unit 102g, separately prepares data to belong to the reference data other than the reference data group, and A verification data group in which data that does not belong to the data is mixed is created, the Mahalanobis distance is calculated for the verification data group, and the reference data group and the other groups are calculated using the threshold set in the threshold setting unit 102f. It is verified whether it can be correctly determined (step SA-7).
[0170]
That is, the judgment accuracy verification unit 102g accesses the DNA microarray measurement data file 106a and the like, separately prepares data to be included in the reference data (patient data to which the result of IFN administration is added) in addition to the reference data group, and Data that does not belong to the reference data is mixed, and this is used as a verification data group.
[0171]
Then, the judgment accuracy verification unit 102g stores the verification data group in a predetermined storage area of the verification data group file 106g.
[0172]
Then, the judgment accuracy verification unit 102g calculates the Mahalanobis distance for the verification data group in the same manner as in step SA-5, and then uses the threshold value obtained in step SA-6 to divide the reference data group and the other groups. Verify that it can be determined correctly.
[0173]
Here, as a verification method, there is a verification method using a discrimination rate. However, in the MT system method, a ratio of a distance obtained from input data to an expected measurement result is represented by an SN ratio (signal-noise ratio). It is common to verify the accuracy.
[0174]
For example, a data group that is found to be effective for interferon administration other than the reference data is within the threshold, and that the other populations are outside the threshold by verifying the SN ratio that the interferon administration is effective. And the linear relationship between the calculated Mahalanobis distances. For the item selection by the orthogonal table and the S / N ratio, refer to the reference document (“2.5 item selection” on page 19 or “6.6 MT system” on page 62 of “Technical development in MT system” which is the above-mentioned reference document). Parameter design ”).)
[0175]
(8) Review
If it is determined in step SA-7 that sufficient accuracy cannot be obtained, the examination of the item selected as the analysis data item in step SA-3 and the reference data group selected in step SA-2 was insufficient. It is possible.
[0176]
Therefore, the DNA microarray data analyzer 100 accesses the DNA microarray measurement data file 106a to the analysis data item 106d by the processing of the review unit 102h, and determines that the determination accuracy verification unit 102g cannot obtain sufficient accuracy. If so, the reference data group selected by the data group selection unit 102b and / or the analysis data item determined by the analysis data item determination unit 102c is reviewed (step SA-8).
[0177]
Here, the reexamination unit 102h verifies the standard data group by the processing of the standard data reexamination unit 102j using the standard deviation of each item in the standard data group and the frequency of appearance of the data based on the degree of dispersion of the standard data. I do.
[0178]
There are several methods for reselecting items. For example, a method of referring to the opinions of experts on data to be handled, a method of selecting data having a large average difference between a reference data group and other groups, and an orthogonal table proposed by the MT system method. A method of verifying the validity of each item by the SN ratio in a plurality of combinations in which items are randomly selected using the method is exemplified.
[0179]
For example, the reviewing unit 102h verifies which spot (gene expression status) of the DNA microarray to use in the analysis by the analysis data item reviewing unit 102k in a plurality of combinations using an orthogonal table. After the evaluation with the SN ratio, the item whose SN ratio increases when used as an item is used as the item.
[0180]
(9) Confirm
As a result of the verification in step SA-7, when it is determined that sufficient accuracy can be ensured, the DNA microarray data analysis device 100 determines the database (unit space) and the threshold by the processing of the determination unit 102i (step SA-7). -9).
[0181]
For example, as a result of verification, it is determined at the stage where it is determined that the effectiveness of the administration of interferon can be sufficiently diagnosed from a medical standpoint based on the input data.
[0182]
(10) Start classification
After the database (unit space) is determined, it is used to calculate Mahalanobis distance for unknown data, express the properties of the data by distance, and use it for diagnosis and prediction of various data (step SA-10). .
[0183]
For example, when the effectiveness of interferon administration is verified by a DNA microarray and the MT system method, Mahalanobis distance is calculated in a unit space database created in advance using expression status data of the DNA microarray + specimen information as input data, and determined in advance. Using the threshold of Mahalanobis distance, we diagnose the effectiveness and predict the efficacy of interferon administration.
Thus, the DNA microarray analysis processing by the MT system is completed.
[0184]
[Other embodiments]
Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, but may be applied to various different embodiments within the scope of the technical idea described in the claims. It may be implemented.
[0185]
For example, the case where the DNA microarray data analysis apparatus 100 performs the processing in a stand-alone form has been described as an example, but the processing is performed in response to a request from a client terminal configured in a separate housing from the DNA microarray data analysis apparatus 100. Then, the processing result may be returned to the client terminal.
[0186]
Further, of the processes described in the embodiment, all or a part of the processes described as being performed automatically may be manually performed, or all of the processes described as being performed manually may be performed. Alternatively, a part thereof can be automatically performed by a known method.
In addition, the processing procedures, control procedures, specific names, information including parameters such as various registration data and search conditions, screen examples, and database configurations shown in the above-described documents and drawings, except where otherwise noted, It can be changed arbitrarily.
[0187]
Further, regarding the DNA microarray data analyzer 100, the components shown in the drawings are functionally conceptual, and need not necessarily be physically configured as shown in the drawings.
[0188]
For example, all or any part of the processing functions provided in each unit or each device of the DNA microarray data analysis device 100, particularly each processing function performed by the control unit 102, are replaced with a CPU (Central Processing Unit) and the CPU. And can be realized as hardware by wired logic. The program is recorded on a recording medium described later, and is mechanically read by the DNA microarray data analyzer 100 as needed.
[0189]
That is, a computer program for giving instructions to the CPU in cooperation with an OS (Operating System) and performing various processes is recorded in the storage unit 106 such as a ROM or an HD. This computer program is executed by being loaded into a RAM or the like, and configures the control unit 102 in cooperation with the CPU. Further, this computer program may be recorded in an application program server connected to the DNA microarray data analyzer 100 via an arbitrary network 300, and may be downloaded in whole or in part as necessary. It is possible.
[0190]
Further, the program according to the present invention can be stored in a computer-readable recording medium. Here, the “recording medium” refers to an arbitrary “portable physical medium” such as a flexible disk, a magneto-optical disk, a ROM, an EPROM, an EEPROM, a CD-ROM, an MO, a DVD, and the like, and a built-in various computer systems. A short-term program such as a communication line or a carrier wave when transmitting the program via an arbitrary "fixed physical medium" such as ROM, RAM, HD, or a network represented by LAN, WAN, or the Internet. "Communications medium" that holds.
[0191]
The “program” is a data processing method described in an arbitrary language or description method, and may be in any format such as a source code or a binary code. The “program” is not necessarily limited to a single program, but may be distributed in the form of a plurality of modules or libraries, or may operate in cooperation with a separate program represented by an OS (Operating System). Includes those that achieve functions. Note that a known configuration and procedure can be used for a specific configuration, a reading procedure, an installation procedure after reading, and the like in each apparatus described in the embodiments.
[0192]
Various databases and the like (DNA microarray measurement data file 106a to verification data group file 106g) stored in the storage unit 106 include memory devices such as RAM and ROM, fixed disk devices such as hard disks, and storage means such as flexible disks and optical disks. And stores various programs, tables, files, databases, files for web pages, and the like used for various processes and for providing a website.
[0193]
Further, the DNA microarray data analyzer 100 connects a peripheral device such as a printer, a monitor, or an image scanner to an information processing device such as an information processing terminal such as a known personal computer or a workstation, and connects the information processing device of the present invention to the information processing device. The method may be implemented by implementing software (including programs, data, and the like) for implementing the method.
[0194]
Further, the specific form of the dispersion / integration of the DNA microarray data analyzer 100 and the like is not limited to those shown in the description and the drawings, and all or a part thereof may be functionally divided into arbitrary units corresponding to various loads and the like. Alternatively, they can be physically distributed and integrated (for example, grid computing). For example, each database may be independently configured as an independent database device, or a part of the processing may be realized using a CGI (Common Gateway Interface).
[0195]
The network 300 has a function of interconnecting the DNA microarray data analyzer 100 and the external system 200, and includes, for example, the Internet, an intranet, a LAN (including both wired / wireless), a VAN, Personal computer communication network, public telephone network (including both analog and digital), leased line network (including both analog and digital), CATV network, IMT2000 system, GSM system, PDC / PDC-P system, etc. Or a local radio network such as Bluetooth, a PHS network, or a satellite communication network such as CS, BS or ISDB. That is, the present system can transmit and receive various data via any network regardless of wired or wireless.
[0196]
【The invention's effect】
As described above in detail, according to the present invention, since the measurement data of the DNA microarray is analyzed using the Mahalanobis-Taguchi system, a large amount of gene data is analyzed at high speed and with high accuracy by applying the MT system method. The present invention can provide a DNA microarray data analysis method, a DNA microarray data analysis device, a program, and a recording medium that can be used.
[0197]
That is, in the conventional DNA microarray data analysis method using a neural network that requires teacher data, the features of all the teacher data (data for which the determination result is clear, such as the sensitivity of a clinically determined drug) are quantified. However, in the MT system method, the analysis can be started with reference to the vague contents such as whether or not the drug has an effect and whether or not the drug belongs to a specific group. Teacher data) Preparation becomes very simple.
[0198]
In the conventional DNA microarray data analysis method using neural network learning, multiple regression analysis coefficient calculation, and the like, the same operation is repeated a plurality of times to form neural circuits and parameters in an optimal state. In the method, a database (reference data) can be generated by one calculation and can be processed at a high speed, so that the operation time can be extremely shortened.
[0199]
Further, in the conventional DNA microarray data analysis method using a neural network or the like, the calculation process is treated as a black box, and the calculation process of the obtained result is not clear (that is, the validity of the calculation process cannot be verified). Although several calculation methods have been proposed in the MT system method, they are all open calculation formulas, and it is clear that the calculation formula focuses on the correlation of all reference data. Can be guaranteed.
[0200]
In the conventional DNA microarray data analysis method using a neural network or the like, the calculated result fluctuates depending on the contents of the teacher data and the software used, and there is no reproducibility when reconstructing a neural circuit. In the method, the calculation is performed using a published calculation formula, so that the same result can be always obtained by using the same data, and the reproducibility can be extremely improved.
[0201]
Further, in a DNA microarray data analysis method using a general clustering technique, a difference between a specimen and a gene is expressed as a relative one within a group by dividing the data into groups within a given data group. In the MT system method, once a unit space database is generated, it can be expressed as an absolute value by expressing it as a distance from the origin of this reference, and other than the analyzed data group, the Mahalanobis distance from the unit space database can be expressed. Can be evaluated simply by calculating the evaluation result, so that the evaluation result becomes stable every time.
[0202]
Furthermore, in the MT system method, a procedure for mechanically verifying which items (gene expression data and specimen information) are effective for discrimination is defined (item selection). The most appropriate items can be selected efficiently without being bound by biased opinions.
[0203]
Further, according to the present invention, the gene to be analyzed is classified from the measurement data of the DNA microarray using the Mahalanobis-Taguchi system, so that a large amount of gene data can be classified at high speed and with high accuracy by applying the MT system method. , A DNA microarray data analysis device, a program, and a recording medium.
[0204]
According to the present invention, a gene to be analyzed is selected from the measurement data of the DNA microarray of the sample using the Mahalanobis-Taguchi system, and the clinical sample is classified based on the selected gene data using the Mahalanobis-Taguchi system. Therefore, by applying the MT system method, it is possible to provide a DNA microarray data analysis method, a DNA microarray data analysis device, a program, and a recording medium that can classify a large amount of microarray measurement data at high speed and with high accuracy. it can.
[0205]
Further, according to the present invention, a reference data group is selected by collecting reference data according to a discrimination item to be analyzed from the measurement data of the DNA microarray, and input data when discriminating by the MT system method is used. Items to be analyzed in the MT system method, and a unit space database indicating a Mahalanobis space of the reference data included in the selected reference data group is created. Calculate the Mahalanobis distance between the data and the comparison data that does not clearly belong to the reference data, and, based on the calculated Mahalanobis distance between the reference data and the comparison data, one or more thresholds that serve as boundaries between the reference data group and the comparison data group Is set, and the Mahalanobis distance of unknown data is calculated using the unit space database and the selection items, Since the unknown data is classified by making a determination based on the threshold value, a DNA microarray data analysis method, a DNA microarray data analysis device, a program, and a program that can analyze a large amount of genetic data at high speed and with high accuracy by applying the MT system method In addition, a recording medium can be provided.
[0206]
Further, according to the present invention, data to belong to the reference data besides the reference data group is separately prepared, and a verification data group in which the data and the data not belonging to the reference data are mixed is created, and the Mahalanobis distance is obtained for the verification data group. Is calculated, and it is verified whether the reference data group and the other groups can be correctly distinguished using the set threshold value. Therefore, a DNA microarray data analysis method and a DNA microarray data analysis that can perform the discrimination work more efficiently An apparatus, a program, and a recording medium can be provided.
[0207]
Further, according to the present invention, when it is determined that sufficient accuracy cannot be obtained, the selected reference data group and / or the determined analysis data item are re-examined, so that the determination operation can be performed with higher accuracy and It is possible to provide a DNA microarray data analysis method, a DNA microarray data analysis device, a program, and a recording medium that can be performed efficiently.
[0208]
Further, according to the present invention, the degree of variation of the reference data is verified using the standard deviation and the frequency of data appearance of each item in the reference data group, and the reference data having the small degree of variation is used. It is possible to provide a DNA microarray data analysis method, a DNA microarray data analysis device, a program, and a recording medium that can perform a discriminating operation more accurately and efficiently by reviewing data.
[0209]
Furthermore, according to the present invention, which combination of spots on the DNA microarray to use for measurement data is verified using a plurality of combinations using an orthogonal table, evaluated using the S / N ratio, and used as an analysis data item, the S / N ratio increases. A DNA microarray data analysis method, a DNA microarray data analysis device, a program, which can perform discrimination work more accurately and efficiently by re-examining the analysis data items because the data are used as analysis data items. In addition, a recording medium can be provided.
[Brief description of the drawings]
FIG. 1 is a principle configuration diagram showing a basic principle of the present invention.
FIG. 2 is a block diagram illustrating an example of a configuration of the present system to which the present invention is applied.
FIG. 3 is a flowchart illustrating an example of a DNA microarray analysis process by the MT system of the present system in the present embodiment.
FIG. 4 is a diagram illustrating an example of a case where a reference data group and a comparison data group are displayed in a distance histogram and a threshold is set.
FIG. 5 is a diagram showing an example of user information stored in a DNA microarray measurement data file 106a.
FIG. 6 is a block diagram illustrating an example of a configuration of a review unit 102h.
[Explanation of symbols]
100 DNA microarray data analyzer
102 control unit
102a Analysis purpose determination unit
102b Data group selector
102c Analysis data item determination unit
102d Unit space database creation unit
102e Mahalanobis distance calculator
102f threshold setting unit
102g Judgment accuracy verification unit
102h Review section
102i Confirmation unit
102j Reference data review section
102k Analysis data item review section
104 Communication control interface unit
106 storage unit
106a DNA microarray measurement data file
106b Reference data group file
106c Comparison data group file
106d Analysis data item
106e Unit space database
106f Mahalanobis distance file
106g Verification data group file
108 I / O control interface
112 input device
114 Output device
200 External system
300 Network

Claims (19)

A DNA microarray data analysis method, comprising analyzing measurement data of a DNA microarray using a Mahalanobis Taguchi system. A method for analyzing DNA microarray data, comprising classifying genes to be analyzed from DNA microarray measurement data using a Mahalanobis Taguchi system. A DNA microarray characterized in that a gene to be analyzed is selected from measurement data of a sample DNA microarray using a Mahalanobis-Taguchi system, and a clinical sample is classified based on the selected gene data using a Mahalanobis-Taguchi system. Data analysis method. A data group selection step of selecting a reference data group by collecting reference data according to the discrimination item to be analyzed from the measurement data of the DNA microarray;
An analysis data item determining step of determining an analysis data item which is an item to be input data when the Mahalanobis-Taguchi system is used to determine the analysis data item,
A unit space database creation step of creating a unit space database indicating a Mahalanobis space of the reference data included in the reference data group selected in the data group selection step,
Mahalanobis distance of the reference data selected in the data group selection step, Mahalanobis distance calculation step of calculating a Mahalanobis distance by collecting comparison data that does not clearly belong to the reference data,
Threshold setting for setting one or a plurality of thresholds serving as boundaries between the reference data group and the comparison data group based on the Mahalanobis distance of the reference data calculated in the Mahalanobis distance calculation step and the Mahalanobis distance of the comparison data Steps and
Using the unit space database and the selected item, calculate the Mahalanobis distance of the unknown data, and determine the Mahalanobis distance by the threshold, to perform a discriminant analysis of the unknown data,
A DNA microarray data analysis method, comprising: In addition to the reference data group, separately prepare data to belong to the reference data, create a verification data group in which this and data that does not belong to the reference data are mixed, calculate the Mahalanobis distance for the verification data group, A determination accuracy verification step of verifying whether or not the reference data group and the other groups can be correctly determined using the threshold value set in the threshold value setting step,
The DNA microarray data analysis method according to claim 4, further comprising: If it is determined that sufficient accuracy cannot be obtained in the verification step, the reference data group selected in the data group selection step and / or the analysis data item determined in the analysis data item determination step A review step to review the
The DNA microarray data analysis method according to claim 5, further comprising: The above review step
A reference data reexamination step of verifying the degree of variation of the reference data using a standard deviation or a data appearance frequency of each item in the reference data group, and using the reference data having a small degree of variation,
The DNA microarray data analysis method according to claim 6, further comprising: The above review step
Using an orthogonal table to verify which spots of the DNA microarray to use for the measurement data in a plurality of combinations and evaluating the S / N ratio, use the analysis data items that increase the S / N ratio as the analysis data items. Analysis data item review step to be used as
The DNA microarray data analysis method according to claim 6, further comprising: Data group selecting means for selecting a reference data group by collecting reference data corresponding to the discrimination item to be analyzed from the measurement data of the DNA microarray;
Analysis data item determination means for determining an analysis data item which is an item serving as input data when the Mahalanobis-Taguchi system is used to make a determination in the Mahalanobis-Taguchi system,
A unit space database creating means for creating a unit space database indicating a Mahalanobis space of the reference data included in the reference data group selected in the data group selection step,
Mahalanobis distance of the reference data selected in the data group selection step, Mahalanobis distance calculating means for calculating a Mahalanobis distance by collecting comparison data that does not clearly belong to the reference data,
Threshold setting for setting one or a plurality of thresholds serving as boundaries between the reference data group and the comparison data group based on the Mahalanobis distance of the reference data calculated in the Mahalanobis distance calculation step and the Mahalanobis distance of the comparison data Means,
Using the unit space database and the selected item, calculate the Mahalanobis distance of the unknown data, and determine the Mahalanobis distance by the threshold, thereby performing a discriminant analysis of the unknown data,
A DNA microarray data analyzer, comprising: In addition to the reference data group, separately prepare data to belong to the reference data, create a verification data group in which this and data that does not belong to the reference data are mixed, calculate the Mahalanobis distance for the verification data group, Discrimination accuracy verification means for verifying whether the reference data group and the other groups can be correctly determined using the threshold value set by the threshold value setting means,
The DNA microarray data analysis device according to claim 9, further comprising: If the verification means determines that sufficient accuracy cannot be obtained, the reference data group selected by the data group selection means and / or the analysis data item determined by the analysis data item determination means Reexamination means to reexamine
The DNA microarray data analysis device according to claim 10, further comprising: The above review means
A reference data reexamination means for verifying the degree of variation of the reference data using the standard deviation and the frequency of data appearance of each item in the reference data group, and using the reference data having a small degree of variation,
The DNA microarray data analysis device according to claim 11, further comprising: The above review means
Using an orthogonal table to verify which spots of the DNA microarray to use for the measurement data in a plurality of combinations, evaluate them with the SN ratio, and then use the analysis data items that increase the SN ratio when used as the analysis data items. Analysis data item review means to be used as
The DNA microarray data analysis device according to claim 11 or 12, further comprising: A data group selection step of selecting a reference data group by collecting reference data according to the discrimination item to be analyzed from the measurement data of the DNA microarray;
An analysis data item determining step of determining an analysis data item which is an item to be input data when the Mahalanobis-Taguchi system is used to determine the analysis data item,
A unit space database creation step of creating a unit space database indicating a Mahalanobis space of the reference data included in the reference data group selected in the data group selection step,
Mahalanobis distance of the reference data selected in the data group selection step, Mahalanobis distance calculation step of calculating a Mahalanobis distance by collecting comparison data that does not clearly belong to the reference data,
Threshold setting for setting one or a plurality of thresholds serving as boundaries between the reference data group and the comparison data group based on the Mahalanobis distance of the reference data calculated in the Mahalanobis distance calculation step and the Mahalanobis distance of the comparison data Steps and
A classification step of performing a discriminant analysis of unknown data by calculating a Mahalanobis distance of unknown data by using the unit space database and the selected item, and determining the Mahalanobis distance by the threshold value;
A program for causing a computer to execute a DNA microarray data analysis method including: In addition to the reference data group, separately prepare data to belong to the reference data, create a verification data group in which this and data that does not belong to the reference data are mixed, calculate the Mahalanobis distance for the verification data group, A determination accuracy verification step of verifying whether or not the reference data group and the other groups can be correctly determined using the threshold value set in the threshold value setting step,
The program according to claim 14, further comprising: If it is determined that sufficient accuracy cannot be obtained in the verification step, the reference data group selected in the data group selection step and / or the analysis data item determined in the analysis data item determination step A review step to review the
The program according to claim 15, further comprising: The above review step
A reference data reexamination step of verifying the degree of variation of the reference data using a standard deviation or a data appearance frequency of each item in the reference data group, and using the reference data having a small degree of variation,
17. The program according to claim 16, further comprising: The above review step
Using an orthogonal table to verify which spots of the DNA microarray to use for the measurement data in a plurality of combinations and evaluating the S / N ratio, use the analysis data items that increase the S / N ratio as the analysis data items. Analysis data item review step to be used as
The program according to claim 16, further comprising: A computer-readable recording medium having recorded thereon the program according to any one of claims 14 to 18.

Images