JP4568861B2 - Gene expression profile comparison device - Google Patents

Description

  The present invention relates to an apparatus for obtaining a correlation between a gene expression profile of a known cell stored in a database and a gene expression profile inputted as a question.

  Currently, gene expression monitoring using a DNA microarray has been actively studied. For example, Non-Patent Document 1 discloses a study for finding a gene having a difference from gene expression profile data, and Non-Patent Document 2 discloses a study for finding cancer from a gene expression profile.

  If a cell having the profile can be identified from the gene expression profile using a DNA microarray, it is very useful in the fields of pathological analysis, criminal investigation and the like. An unknown cell can be identified by searching a cell that matches a gene expression profile of an unknown cell or a cell having a high degree of similarity from a gene expression profile database covering all cells.

  As a technique for comparing a known gene expression profile with an unknown gene expression profile (question profile), a technique for comparing expression levels of genes included in each gene expression profile is known.

In this method, first, a gene expression profile and a gene included in the gene expression profile are specified, and a query for obtaining the expression level of the gene is transmitted to a database. Then, the response from the database (expression level data) is compared with the expression level of the same gene included in the query profile. By this method, the degree of similarity between the gene expression profile stored in the database and the query profile is obtained by comparing the expression levels of a plurality of genes included in the gene expression profile to be compared.
“Tumorclassification and marker gene prediction by feature selection and fuzzyc-means clustering using microarray data.”, Wang J, Bo TH, Jonassen I, Myklebost O, Hovig E, BMC Bioinformatics. 2003 Dec 02; 4 (1): 60. “Multiclasscancer diagnosis using tumor gene expression signatures.”, Ramaswamy S, TamayoP, Rifkin R, Mukherjee S, Yeang CH, Angelo M, Ladd C, Reich M, Latulippe E, Mesirov JP, Poggio T, Gerald W, Loda M, Lander ES, Golub TR, Proc Natl Acad SciU S A. 2001 Dec 18; 98 (26): 15149-54.

  However, the above method has a problem that when the gene expression profile to be compared is increased, the calculation process takes a long time and requires a large memory area. In the above method, the query transmitted to the database for obtaining the expression level of the gene uses a general-purpose function, which is convenient, but it takes time to respond. In addition, when the data of the designated gene is missing, the time required for inquiring the expression level of the gene is wasted. For these reasons, the calculation process takes time.

  In view of the above background, an object of the present invention is to provide a gene expression profile comparison apparatus capable of comparing gene expression profile data at high speed.

  The gene expression profile comparison apparatus of the present invention includes a gene expression profile database storing gene expression profiles, input means for receiving input of a comparison target question profile, the question profile input by the input means, and the gene expression profile Calculating means for calculating a correlation with the gene expression profile stored in the database, wherein the calculating means determines an identifier of the gene constituting the platform of the gene expression profile stored in the gene expression profile database by the identifier. The first constituent gene sequence data is generated by storing in a continuous memory area according to a predetermined order, and the identifiers of genes constituting the query profile are stored in the continuous memory area according to the order determined by the identifiers. The second constituent gene sequence data is generated, the gene identifiers are read in order from the top of the first constituent gene sequence data, and the gene identifiers are read in order from the top of the second constituent gene sequence data. When the identifiers of the genes are compared and the identifiers of both genes match, the gene expression profile is stored as a common gene in the platform and the query profile, and the first constituent gene sequence data and the second constituent gene are stored. If the identifier of the next gene is read from the sequence data and compared, and the identifiers of both genes do not match, the identifier of the next gene is read from the sequence data from which the identifier of the earlier gene identifier is read Are read out and compared again, the first constituent gene sequence data and the second constituent gene By repeating to read all the data, either the column data to extract the common genes, determining the correlation by comparing the expression level of the extracted common gene.

  As described above, the first constituent gene sequence data storing the identifiers of the genes constituting the gene expression profile platform in the order of the identifiers and the second constituent gene sequence data storing the identifiers of the genes constituting the query profile in the order of the identifiers are generated. To do. Thereby, the identifier of the gene which comprises a platform can be read according to a predetermined order (for example, small order) by the simple process of reading the 1st constituent gene arrangement data in order from the head. Similarly, the identifier of the gene constituting the question profile can be read in a predetermined order by a simple process of reading the second constituent gene sequence data in order from the top. Since the identifiers of the genes are read from each of the first constituent gene sequence data and the second constituent gene sequence data according to a predetermined order (for example, in ascending order), the order of the identifiers is determined when the two read genes do not match. It can be determined that the earlier gene has no gene common to the other sequence data. Accordingly, the next gene (the gene with the next highest identifier order) is read from the sequence data from which the gene with the earlier identifier order is read, and the comparison process is repeated. By repeating this process, it can be determined at high speed whether or not there is a common gene without comparing it with all the genes of the other sequence data. By comparing the expression levels of the common genes obtained in this way, it is possible to speed up the calculation processing for obtaining the correlation between the gene expression profile data and the query profile.

  In the gene expression profile comparison device, the computing means stores the identifiers of the genes included in the gene expression profiles stored in the gene expression profile database in successive memory areas according to the expression order of the respective genes. Gene expression data of one expression order is generated, the data is read in order from the beginning of the first expression order gene sequence data, and when the read gene is the common gene, the read gene is the common number By determining whether the gene is a gene, the expression rank among the common genes included in the gene expression profile is obtained, and the identifier of the gene included in the query profile is stored in a continuous memory according to the expression rank of each gene. Generating the second expression sequence gene sequence data stored in the region, If the read-out gene is the common gene, the query profile is identified by specifying the number of the common gene when the read-out gene is the common gene. The expression order among the common genes contained in may be determined.

  In this way, the common gene is read from the head of the first expression sequence gene sequence data in which the identifiers of the genes included in the gene expression profile are stored in accordance with the expression order, and the order is assigned in the order of reading, so that the gene expression profile is obtained. The expression order among the contained common genes can be determined. Similarly, the order of expression among the common genes included in the query profile can be determined by simple processing.

  In the gene expression profile comparison apparatus, the computing unit can assign consecutive identification numbers to genes that can be compared as the identifier, and write expression rank data for the number of genes to which the identification number is assigned. A memory area for the first expression rank sequence data and the second expression rank sequence data is secured, and the expression rank in the common gene included in the gene expression profile is expressed by the first expression rank sequence data. The common gene identification number is written in the position indicated by the identification number of each common gene, and the expression rank in the common gene included in the query profile is expressed as the identification number of each common gene in the second expression rank sequence data. The first expression rank sequence data and the second expression rank sequence data Reads data of expression rank in the corresponding position in the data, based on the difference between the read expression rank may be calculated rank correlation coefficient between the gene expression profile and questions profile.

  Thus, by assigning consecutive identification numbers to genes that can be compared, and managing the gene information with the sequence data written at the position indicated by the identification number, the gene information can be acquired at high speed. According to the configuration of the present invention, the memory region of the first expression rank sequence data in which the same number of expression rank data as the assigned identification number can be written is secured, and the expression rank data is placed at the position indicated by the identification number of each common gene. By writing, data on the expression order of common genes is written in the order of gene identification numbers. Therefore, the expression rank data of the gene corresponding to the identification number can be read by a simple process of reading data from the position indicated by the identification number of the first expression rank sequence data. Similarly, expression rank data of common genes included in the query profile can be read out by a simple process.

  In the above gene expression profile comparison device, the reading of the expression rank data from the first expression rank sequence data and the second expression rank sequence data is performed in the order of the low expression rank in the second expression rank sequence data, If the expression rank data of the first expression rank sequence data corresponding to the expression rank data read from the second expression rank sequence data is missing, the number of missing values is stored, and the first When calculating the difference between the expression ranks read from the first expression rank sequence data and the second expression rank sequence data, the number of missing values at that time from the expression rank read from the second expression rank sequence data Then, the difference in the expression order may be calculated.

  In gene expression profile data, even if a gene is present on the platform, the expression level data is not always present and may be missing. For the defective gene, comparison with the expression level of the gene in the question profile cannot be performed. According to the present invention, when the expression rank data of the first expression rank sequence data (gene expression profile) corresponding to the expression rank data of the second expression rank sequence data (question profile) is missing, Since expression rank data cannot be compared for a gene, it is excluded from the target of comparison of expression rank data. Since reading is performed in order of increasing expression rank in the second expression rank sequence data, after the missing value is detected by subtracting the number of missing values from the expression rank data read after the missing value is detected. The rank difference is calculated by raising the expression rank data of the number of missing values, and the correlation can be appropriately obtained.

  The correlation calculation device according to the present invention includes a rank database storing rank data indicating ranks assigned to each of a plurality of identifiers, and an input for receiving rank data to be compared with rank data stored in the rank database. Means for calculating the correlation between the rank data input by the input means and the rank data stored in the rank database, and the calculation means adds the rank data stored in the rank database to the rank data stored in the rank database. The identifiers included are stored in a continuous memory area in the order determined by the identifiers to generate the first sequence data, and the identifiers of the genes constituting the rank data input to the input means are determined by the identifiers. According to the order, the second array data is generated by storing in a continuous memory area, and the first array data The identifiers are read in order from the beginning, the identifiers are read in order from the beginning of the second array data, the read identifiers are compared, and if both identifiers match, the matched identifier is stored as a common identifier and the first If the next identifier is read out from the array data and the second array data and compared, and the two identifiers do not match, the next identifier is read out from the array data from which the identifier in the earlier order is read out. The process of reading and comparing again is repeated until all the data of the first array data and the second array data is read, and the common identifier is extracted, and the ranks of the extracted common identifiers are compared with each other. Find the correlation.

  With this configuration, similar to the gene expression profile comparison apparatus of the present invention, it is possible to perform a calculation process for obtaining the correlation between the first sequence data and the second sequence data at high speed. Various configurations of the gene expression profile comparison apparatus of the present invention can also be applied to the correlation calculation apparatus of the present invention.

  The program of the present invention is a computer-readable program that causes a computer to calculate a correlation between a gene expression profile stored in a gene expression profile database and a query profile to be compared. The genes constituting the query profile are generated by storing the identifiers of the genes constituting the platform of the gene expression profile stored in the memory in a continuous memory area according to the order determined by the identifiers and generating the first constituent gene sequence data. Are stored in a continuous memory area according to the order determined by the identifiers to generate second constituent gene sequence data, and the gene identifiers are read in order from the top of the first constituent gene sequence data. The gene identifiers are read in order from the top of the second constituent gene sequence data, the identifiers of the read genes are compared, and when the identifiers of both genes match, the common gene expression profile platform and the query profile are common When the identifier of the next gene is read out from the first constituent gene sequence data and the second constituent gene sequence data and compared, and the identifiers of both genes do not match, The process of reading out the identifier of the next gene from the sequence data from which the identifier of the earlier one is read out and comparing it again is performed using either one of the first component gene sequence data and the second component gene sequence data. Repeat until all the data are read out to extract common genes, and the expression level of the extracted common genes Causing obtains the correlation by comparing.

  With this configuration, similar to the gene expression profile comparison apparatus of the present invention, it is possible to perform a calculation process for obtaining the correlation between the gene expression profile data and the query profile at high speed. Various configurations of the gene expression profile comparison apparatus of the present invention can be applied to the gene expression profile comparison program of the present invention.

  According to the present invention, by reading genes from the first constituent gene sequence data and the second constituent gene sequence data in which constituent genes are arranged in the order of identifiers, the identifiers of the genes are determined from the respective sequence data according to a predetermined order (for example, (In ascending order), common genes can be obtained at high speed with simple processing. By comparing the common genes obtained in this way, there is an excellent effect that it is possible to speed up the calculation processing for obtaining the correlation between the gene expression profile data and the query profile.

  Hereinafter, a gene expression profile comparison apparatus according to an embodiment of the present invention will be described with reference to the drawings.

  FIG. 1 is a diagram illustrating a configuration of a gene expression profile comparison apparatus 10 according to the present embodiment. The gene expression profile comparison apparatus 10 is stored in an input unit 12 that inputs a gene expression profile to be compared as a question profile, and an input question profile and gene expression profile database (hereinafter referred to as “gene expression profile DB”) 16. CPU 14 for comparing the gene expression profile data and an output unit 20 for outputting the comparison result. A memory 18 is connected to the CPU 14. The CPU 14 writes data into the memory 18 and reads data from the memory 18 when performing calculations.

  The gene expression profile comparison apparatus 10 is configured by a personal computer, for example. The input unit 12 includes, for example, a keyboard, a mouse, a CD-ROM reader, and the like. The output unit 20 includes, for example, a monitor, a printer, a CD-ROM writing device, and the like.

  The gene expression profile DB 16 stores a plurality of gene expression profile data. Genes constituting the gene expression profile data are specified by an identification number. An identification number is assigned to a gene handled as a comparison target in the gene expression profile comparison apparatus 10. Here, the genes handled as comparison targets are all genes possessed by humans, for example, when analyzing human gene expression profiles. In addition, when it is difficult to have data for all genes, genes constituting the platform of gene expression profiles stored in the gene expression profile DB 16 may be used as genes to be compared. For example, consecutive identification numbers unique to the gene expression profile comparison device 10 are assigned to the genes handled by the gene expression profile comparison device 10. At this time, a general identifier and an identification number peculiar to the gene expression profile comparison device 10 are converted using a table indicating correspondence between the identification numbers used in the gene expression profile comparison device 10 and the identifiers of commonly used genes. May be.

  FIG. 2A is a diagram illustrating an example of data stored in the gene expression profile DB 16. In the following description, for convenience of explanation, the gene identification number is described as “˜number”, and the expression rank of the gene specified by the identification number is described as “˜rank”. The actual gene expression profile data has information on tens of thousands of genes, but in the present embodiment, for the convenience of explanation, the first to eleventh genes are targeted. The gene expression profile DB 16 stores information on gene expression levels as gene expression ranks. This makes it easy to compare gene expression profiles. In the example shown in FIG. 2 (a), the gene expression profile data is composed of genes No. 7, No. 3, No. 8, No. 1, No. 4, No. 10, No. 11 from the higher expression rank. . That is, the gene expression profile platform stored in the gene expression profile DB 16 does not have the 2nd, 5th, 6th, and 9th genes among all the 1st to 11th genes that can be compared. The gene expression profile DB 16 stores information on genes arranged in accordance with the expression order in FIG. 2A (portions surrounded by bold lines). In addition, the gene expression profile comparison apparatus 10 can sort the information of the gene expression profile data sorted according to the expression rank as shown in FIG. 2A in the order of the identification numbers as shown in FIG.

  Next, a process of comparing the gene expression profile stored in the gene expression profile DB 16 with the question profile by the gene expression profile comparison apparatus 10 of the present embodiment will be described.

First, an example of a question profile input from the input unit 12 will be described.
FIG. 3 is a diagram illustrating an example of a question profile. Similar to the gene expression profile, the question profile is information indicating the expression level of the gene in the order of expression. The input unit 12 receives input of gene information (portions surrounded by bold lines) arranged according to the expression order.

  The gene expression profile comparison apparatus 10 obtains a correlation between the gene expression profile stored in the gene expression profile DB 16 and the question profile input from the input unit 12. First, an outline of the calculation will be described.

FIG. 4 is a flowchart showing calculation processing by the gene expression profile comparison apparatus 10. First, a gene commonly included in the gene expression profile and the query profile is obtained as a common gene (S10). Next, the expression rank in the common gene is obtained (S12), and the rank correlation coefficient is obtained based on the difference in the expression rank in the same common gene (S14).
Hereinafter, the data shown in FIG. 2A and FIG. 3 will be described in detail as an example.

(How to find common genes)
The CPU 14 of the gene expression profile comparison apparatus 10 performs a process of obtaining a common gene between the gene included in the gene expression profile platform stored in the gene expression profile DB 16 and the gene included in the question profile.

  The CPU 14 sorts the identification numbers of the genes included in the platform according to the identification numbers and stores them in the memory 18 as sequence data. This sequence data is referred to as first constituent gene sequence data. FIG. 5A shows first constituent gene sequence data in which gene expression profile data (FIG. 2A) is arranged in the order of identification numbers. Similarly, the identification numbers of the genes included in the question profile are sorted according to the identification number and stored in the memory 18 as sequence data. This sequence data is referred to as second constituent gene sequence data. FIG. 5B shows second constituent gene sequence data in which the query profiles (FIG. 3) are arranged in the order of identification numbers.

  The CPU 14 obtains a common gene by comparing data read in order from the beginning from each of the first component gene sequence data and the second component gene sequence data. Specifically, first, data is read from the top of the array data of the platform (FIG. 5A), and data with an identification number “No. 1” is acquired. Data is read from the beginning of the array data of the question profile (FIG. 5B), and data with an identification number of “No. 1” is acquired. Comparing both the read data, it is found that the “No. 1” gene is common between the platform and the question profile because both are matched with “No. 1”. The CPU 14 stores “No. 1” as a common gene.

  Next, the CPU 14 reads data from the second of the array data of the platform (FIG. 5A), and acquires data having an identification number “No. 3”. Data is read from the second of the array data of the question profile (FIG. 5B), and data with an identification number of “No. 2” is acquired. When the read data is compared, the two data do not match, so the next (third) data is read from the query profile array data (FIG. 5B) from which the smaller identification number is read. The data of the identification number “No. 4” is acquired. Since the data read from the array data of the platform is “No. 3” and the data read from the array data of the question profile is “No. 4”, the two data do not match. The CPU 14 reads the next (third in the array) data from the array data (FIG. 5A) of the platform from which the smaller identification number is read out, and obtains data with the identification number “4”. Comparing the read data, since both match with “No. 4”, it is understood that the “No. 4” gene is common to the platform and the question profile. The CPU 14 stores “No. 4” as a common gene.

  Hereinafter, similarly, the process of reading and comparing the data in order from each of the first constituent gene sequence data and the second constituent gene sequence data is performed for all of the first constituent gene sequence data or the second constituent gene sequence data. The data is read until the common gene is obtained. In this example, the first, fourth, seventh, eighth and eleventh genes are obtained as common genes. In addition, it is preferable to obtain | require the common gene which is common in the gene which comprises a platform, and a query profile before comparing with each gene expression profile memorize | stored in gene expression profile DB16.

  FIG. 6 is a diagram showing an example of sequence data indicating a common gene (hereinafter referred to as “common gene sequence data”). The common gene sequence data stores data on genes 1 to 11 which are comparison targets in the present embodiment. The common gene sequence data has a size capable of storing 11 pieces of data, and the data for the gene with the identification number a is stored in the a-th common gene sequence data. In the common gene sequence data, the value of “TRUE” is stored in the common gene, and the value of “FALSE” of genes other than the common gene is stored. For example, since No. 1 is a common gene, the value of “TRUE” is stored in the first of the sequence data. Since No. 2 is not a common gene, the value of “FALSE” is stored in the second of the sequence data. By storing such common gene sequence data in the memory 18, whether or not the gene is a common gene can be easily checked from the gene identification number. For example, when checking whether or not the a-th gene is a common gene, it can be determined at high speed by reading the a-th data of the sequence data and determining whether it is TRUE or FALSE. In an actual program, by using a gene identification number as a subscript of sequence data, it is possible to refer to the gene information of that identification number.

(How to determine the order of expression among common genes)
Next, a method for obtaining the expression order among common genes will be described.
As shown in FIG. 7A, the CPU 14 generates sequence data in which gene expression profile data is arranged in the order of expression. This sequence data is referred to as expression sequence gene sequence data. In this sequence data, regardless of whether it is a common gene or not, an expression rank is given to genes included in the gene expression profile. The CPU 14 reads the common genes in order from the top of the gene sequence data in the order of expression, assigns the ranks in the order of reading, and obtains the expression rank among the common genes included in the gene expression profile. A configuration for reading the common gene from the gene expression profile data and storing the read expression order of the common gene will be described in detail.

  As shown in FIG. 7C, the CPU 14 generates array data in which the same number of data as the identification number can be written. All the values of the array data are initially set to “0”. This sequence data is referred to as expression rank sequence data. FIG. 7B is the common gene sequence data described above. CPU14 reads the data of an identification number from the head of expression sequence gene sequence data. Here, an identification number “7” is read out. The CPU 14 refers to the value of the common gene sequence data at the position indicated by the identification number. Referring to FIG. 7B, the seventh value of the common gene sequence data is “TRUE”. Therefore, the CPU 14 determines that “No. 7” is a common gene, counts up the expression rank (first rank at this stage), and sets it as the expression rank among the common genes. The CPU 14 writes the obtained expression rank at the position indicated by the identification number in the expression rank sequence data (FIG. 7C).

  Next, the CPU 14 reads data next to the expression sequence gene sequence data (second data). Here, the identification number “No. 3” is read out. Referring to the value of the common gene sequence data (FIG. 7B) at the position indicated by this identification number, it is “FALSE”. Therefore, the CPU 14 determines that the “No. 3” gene is not a common gene, and proceeds to the next process. The CPU 14 reads data next to the expression sequence gene sequence data (third data). Here, an identification number “8” is read out. Referring to the common gene sequence data (FIG. 7B) at the position indicated by this identification number, it is “TRUE”. Therefore, the CPU 14 determines that “No. 8” is a common gene, counts up the expression rank (second rank at this stage), and sets it as the expression rank among the common genes. The CPU 14 writes the obtained expression rank at the position indicated by the identification number in the expression rank sequence data (FIG. 7C). The above process is repeated until all the data in the expression sequence gene sequence data (FIG. 7A) is read, and the expression rank in the common gene is obtained. FIG. 7 (d) is a diagram showing expression rank sequence data in which expression ranks among common genes are stored. “0” indicates that the gene is not a common gene or there is no expression level data. Similarly, CPU14 calculates | requires the expression order in the common gene of a question profile.

(How to calculate rank correlation coefficient)
Next, how to obtain the rank correlation coefficient will be described.
FIG. 8A is a diagram showing the first expression rank sequence data indicating the expression rank in the common gene of the gene expression profile data, and FIG. 8B is the expression rank in the common gene of the query profile. It is a figure which shows the 2nd expression order arrangement | sequence data shown. The memory 18 stores the expression rank sequence data shown in FIGS. 8A and 8B.

第１の発現順位配列データと第２の発現順位配列データからのデータ読み出しは、第２の発現順位配列データにおいて発現順位の高い方から順に読み出す。この例では、７番、１番、８番、１１番、４番の順で発現順位データを読み出す。これにより、次に説明するように、遺伝子発現プロファイルデータに欠損がある場合にも、適切に順位相関係数を求めることができる。 The CPU 14 reads the expression rank data having the same identification number from the first expression rank sequence data and the second expression rank sequence data, and obtains a rank correlation coefficient based on the difference between the read expression rank data. The CPU 14 obtains the rank correlation coefficient r according to the equation (1), where the number n of each common gene and the difference between the ranks are Di.

Data read from the first expression rank sequence data and the second expression rank sequence data is read in order from the highest expression rank in the second expression rank sequence data. In this example, the expression rank data is read in the order of No. 7, No. 1, No. 8, No. 11, No. 4. As a result, as will be described next, the rank correlation coefficient can be appropriately obtained even when the gene expression profile data is deficient.

(How to determine the rank correlation coefficient when there are missing values)
FIG. 9A is a diagram showing an example in which the data of the gene with the identification number “No. 1” is missing in the first expression rank sequence data shown in FIG. As shown in FIG. 9 (a), the “No. 1” gene has no data due to deletion, and thus has a value of “0”. Along with this, the expression ranks in the common genes of “No. 4” and “No. 11” are raised one by one.

  First, the CPU 14 reads out the expression rank data of “No. 7” having the highest rank in the second expression rank sequence data from the first expression rank sequence data and the second expression rank sequence data. Both the expression rank data can be obtained from “No. 7” of the first expression rank sequence data and the second expression rank sequence data, and the CPU 14 obtains the rank difference Di.

  Next, the CPU 14 reads the expression rank data of “No. 1” having the second highest rank in the second expression rank sequence data from the first expression rank sequence data and the second expression rank sequence data. Since “0” indicating no data is read from the first expression rank sequence data, the CPU 14 does not calculate the rank difference Di and counts up the number of missing values. At this stage, the number of missing values is one.

Next, the CPU 14 reads the expression rank data of “# 8” having the third highest rank in the second expression rank sequence data from the first expression rank sequence data and the second expression rank sequence data. Since both the expression rank data can be obtained from “No. 8” of the first expression rank sequence data and the second expression rank sequence data, the rank difference Di is obtained. At this time, the number of missing values is subtracted from the expression rank data read out from the second expression rank sequence data. Here, since the expression rank data “3rd place” is read from “8” in the second expression rank sequence data, the expression rank is set to “2nd place” by subtracting the number of missing values. After that, the difference Di from the expression order read from the first expression order sequence data is obtained. Thereafter, the same processing is repeated to obtain the rank difference Di for the common gene, and the rank correlation coefficient is calculated.
The processing of the CPU 14 in the present embodiment has been described above.

Next, the effect of the gene expression profile comparison apparatus 10 of the present embodiment will be described.
In the gene expression profile comparison apparatus 10 of the present embodiment, the first constituent gene sequence data in which the genes constituting the platform are stored in the order of the identification numbers and the second constituent gene sequence in which the genes constituting the question profile are stored in the order of the identification numbers Data is generated, and gene identification numbers are read and compared in order from the top of each sequence data. Since the gene identification numbers are read from the first component gene sequence data and the second component gene sequence data in ascending order of the identification numbers, common genes can be efficiently searched.

  As a conventional method for obtaining a common gene, a method using a hash method or a tree structure has been known. However, in the hash method, it is necessary to trace a plurality of pointers, and processing takes time. In the hash method, it is necessary to store a frame of a hash table in a memory, and in a method using a tree structure, a pointer representing the tree structure and the like are stored in a memory. On the other hand, according to the method of the present embodiment, since simple array data is used, it is not necessary to follow the pointer, and the memory used is small. Thereby, there is a high possibility that the array data will completely enter the CPU cache, and the calculation process can be speeded up.

  In this embodiment, since data is stored in the sequence in the order of gene identification numbers, the calculation amount can be reduced and the processing speed can be increased. FIG. 10 (a) is a diagram for explaining an example in which a common gene is searched using sequence data in which identification numbers are randomly stored. The gene No. 3 included in the gene expression profile is included in the question profile. An example of searching for whether or not For example, it is necessary to search all data of the question profile until the gene “No. 3” is found. For example, in order to find a gene common to a gene expression profile including 10,000 genes and a query profile including 10,000 genes, 10,000 × 10,000 = 100 million (times) data reading is required. On the other hand, FIG. 10B is a diagram for explaining the search for the common gene according to the present embodiment. In the present embodiment, as shown in FIG. 10B, since the identification numbers are stored in ascending order, it is determined whether or not the “No. 3” gene included in the gene expression profile is included in the query profile. When searching, when the identification number of the gene read from the constituent gene sequence data of the question profile becomes larger than “No. 3”, it can be determined that the “No. 3” gene is not included in the question profile. . Therefore, it is not necessary to search all the data of the question profile, and the common gene can be searched efficiently. This is particularly effective when the number of genes included in the gene expression profile or the query profile is large.

  In addition, when obtaining the expression order among the common genes, the gene sequence data of the identification numbers stored in the order of expression is generated, the gene identification numbers are read out in order from the beginning of the gene sequence data in the order of expression, and read. When genes are common genes, the order of expression is assigned. Therefore, the order of expression among the common genes can be determined at high speed by a simple process of sequentially reading the data from the sequence data.

  In determining whether or not a gene is a common gene, the common gene sequence data storing information indicating whether or not the gene of the identification number is common is referenced at the position specified by the identification number. Whether or not the numbered gene is a common gene can be determined at high speed. That is, the memory 18 storing the common gene sequence data stores information on whether or not the gene is a common gene according to the identification number. By referring to the data at the indicated position, it can be read immediately whether or not it is a common gene.

  Further, since the expression rank sequence data in which the gene expression rank data are arranged in the order of the identification number is used when obtaining the correlation rank coefficient, the expression rank of a predetermined gene can be quickly determined. FIG. 11 is a diagram showing an example of a conventional method for reading the expression order of a predetermined gene. In the conventional apparatus, the expression rank data is stored in a database or the like in association with the identification number. When an identification number is specified and a function for calculating the expression rank is sent to the database, the database searches for a gene that matches the specified identification number, and when the specified gene is found, the expression rank associated with that gene is searched. Go to read data. That is, the address of the area where the expression rank data is recorded is added to the identification number information, and the expression rank data is read with reference to this address. Therefore, it takes time to search for the expression order of a predetermined gene. According to the present embodiment, since the data stored at the position designated by the identification number in the expression rank sequence data is only read, the expression rank data can be acquired in a very short time.

  Further, when calculating the rank correlation coefficient, the expression rank is read in the order of expression of the second expression rank sequence data in which the expression rank data among the common genes included in the query profile is stored in the order of the identification number, and the corresponding expression rank If the data is missing, the missing value is counted up. Then, when calculating the rank correlation coefficient, the number of missing values at that time is subtracted from the expression rank read out from the second expression rank sequence data to obtain a new expression rank. As a result, the number of missing values higher than the read gene identification number is subtracted from the expression rank, and the expression rank among the common genes excluding the defective gene can be calculated. Even when there are missing values, the rank correlation coefficient can be obtained appropriately.

  The gene expression profile comparison apparatus 10 of the present invention has been described in detail with reference to the embodiment, but the present invention is not limited to the above-described embodiment.

  In the above embodiment, a comparison device that compares gene expression profiles has been described, but the present invention can also be applied to a search device that searches a gene expression profile having the highest correlation with a query profile from a plurality of gene expression profiles.

  In the above embodiment, the gene expression profile comparison apparatus 10 has been described. However, the present invention includes a program for executing the processing of the CPU 14 described above.

  In the above-described embodiment, the gene expression profile comparison apparatus 10 that compares gene expression profiles has been described. However, the present invention can also be applied to a correlation calculation apparatus that compares general rank data.

  As described above, the present invention has a superior effect of being able to perform high-speed calculation processing for obtaining a correlation between gene expression profile data and a query profile, and compares gene expression profiles with a gene expression profile comparison apparatus. Useful as such.

It is a figure which shows the structure of the gene expression profile comparison apparatus of embodiment. (A) is a figure showing an example of data memorized by gene expression profile DB. (B) is a diagram showing an example of data obtained by sorting gene expression profile data in the order of identifiers. It is a figure which shows the example of the data of a question profile. It is a figure which shows the calculation process of the rank correlation coefficient by a gene expression profile comparison apparatus. (A) is a figure showing an example of gene expression profile data. (B) is a figure showing an example of a question profile. It is a figure which shows the example of common gene arrangement | sequence data. (A) is a figure showing an example of gene expression profile data. (B) is a figure which shows the example of common gene arrangement | sequence data. (C) is a figure which shows the arrangement | sequence data which can write in the same number of data as an identification number. (D) is a figure which shows the expression order sequence data which memorize | stored the expression order in the common gene. (A) is a figure showing an example of expression rank sequence data of gene expression profile data. (B) is a figure which shows the example of the expression order arrangement | sequence data of a question profile. (A) is a figure showing an example of expression rank sequence data of gene expression profile data. (B) is a figure which shows the example of the expression order arrangement | sequence data of a question profile. (A) is a figure explaining the search of the conventional common gene. (B) is a figure explaining the search of a common gene by this Embodiment. It is a figure which shows the example of the conventional method which reads the expression order of a predetermined gene.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Gene expression profile comparison apparatus 12 Input part 14 CPU
16 Gene expression profile DB
18 Memory 20 Output section

Claims (3)

A gene expression profile database storing gene expression profiles;
An input means for receiving input of a question profile to be compared;
Calculating means for calculating a correlation between the query profile input by the input means and the gene expression profile stored in the gene expression profile database,
The computing means is
Gene identifiers constituting the gene expression profile platform stored in the gene expression profile database are stored in a continuous memory area according to an order determined by the identifiers, and first component gene sequence data is generated,
Storing identifiers of genes constituting the query profile in a continuous memory region according to an order determined by the identifiers to generate second constituent gene sequence data;
Read the identifier of the gene in order from the beginning of the first constituent gene sequence data and read the identifier of the gene in order from the beginning of the second constituent gene sequence data, compare the identifier of the read gene,
When the identifiers of both genes match, the gene expression profile is stored as a common gene common to the platform and the query profile, and the next gene from the first component gene sequence data and the second component gene sequence data is stored. When the identifier is read and compared, and the identifiers of both genes do not match, the identifier of the next gene is read from the sequence data from which the earlier identifier of the genes is read and compared again Is repeated until all the data of either the first constituent gene sequence data and the second constituent gene sequence data are read, and a common gene is extracted,
Gene identifiers included in gene expression profiles stored in the gene expression profile database are stored in a continuous memory region according to the expression order of each gene to generate first expression sequence gene sequence data,
Reading the data in order from the beginning of the first expression order gene sequence data, if the read gene is the common gene, by specifying the number of the common gene the read gene, Find the expression rank among the common genes included in the gene expression profile,
Gene identifiers included in the query profile are stored in a continuous memory region according to the expression order of each gene to generate second expression sequence gene sequence data,
Reading the data in order from the top of the second expression sequence gene sequence data, if the read gene is the common gene, by specifying the number of the common gene the read gene, Find the expression rank among the common genes included in the question profile,
As the identifier, a continuous identification number is given to genes that can be compared,
Securing a memory area for the first expression rank sequence data and the second expression rank sequence data in which the expression rank data for the number of genes assigned with the identification number can be written;
Write the expression rank among the common genes included in the gene expression profile at the position indicated by the identification number of each common gene in the first expression rank sequence data,
Write the expression rank among the common genes included in the query profile at the position indicated by the identification number of each common gene in the second expression rank sequence data,
From the first expression rank sequence data and the second expression rank sequence data, the expression rank data at the corresponding position in each sequence data is read, and based on the read difference in the expression rank, the gene expression profile and A gene expression profile comparison apparatus for obtaining the correlation by calculating a rank correlation coefficient with a question profile. Reading out the expression rank data from the first expression rank sequence data and the second expression rank sequence data is performed in order of increasing expression rank in the second expression rank sequence data,
If the expression rank data of the first expression rank sequence data corresponding to the expression rank data read from the second expression rank sequence data is missing, the number of missing values is stored,
When calculating the difference between the expression ranks read from the first expression rank sequence data and the second expression rank sequence data, the missing value at that time is calculated from the expression rank read from the second expression rank sequence data. The gene expression profile comparison apparatus according to claim 1, wherein the difference in expression rank is calculated after subtracting the number of the expression numbers. A computer-readable program for causing a computer to calculate a correlation between a gene expression profile stored in a gene expression profile database and a comparison target query profile,
In the computer,
Gene identifiers constituting the gene expression profile platform stored in the gene expression profile database are stored in a continuous memory area according to an order determined by the identifiers, and first component gene sequence data is generated,
Storing identifiers of genes constituting the query profile in a continuous memory region according to an order determined by the identifiers to generate second constituent gene sequence data;
Read the identifier of the gene in order from the top of the first constituent gene sequence data and read the identifier of the gene in order from the top of the second constituent gene sequence data, and compare the identifier of the read gene,
When the identifiers of both genes match, the gene expression profile is stored as a common gene common to the platform and the query profile, and the next gene from the first component gene sequence data and the second component gene sequence data is stored. When the identifier is read and compared, and the identifiers of both genes do not match, the identifier of the next gene is read from the sequence data from which the earlier identifier of the genes is read and compared again Is repeated until all the data of either the first constituent gene sequence data and the second constituent gene sequence data is read, and the common gene is extracted,
Gene identifiers included in gene expression profiles stored in the gene expression profile database are stored in a continuous memory region according to the expression order of each gene to generate first expression sequence gene sequence data,
Reading the data in order from the beginning of the first expression order gene sequence data, if the read gene is the common gene, by specifying the number of the common gene the read gene, Let the expression rank among the common genes included in the gene expression profile be determined,
Gene identifiers included in the query profile are stored in a continuous memory area according to the expression order of each gene to generate second expression sequence gene sequence data,
Data is read in order from the beginning of the second expression order gene sequence data, and when the read gene is the common gene, by specifying the number of the common gene the read gene, Let the expression rank among the common genes included in the question profile,
As the identifier, a continuous identification number is given to genes that can be compared,
Securing a memory area for the first expression rank sequence data and the second expression rank sequence data in which the expression rank data for the number of genes assigned with the identification number can be written;
Writing the expression rank among the common genes included in the gene expression profile at the position indicated by the identification number of each common gene in the first expression rank sequence data;
Writing the expression rank among the common genes included in the query profile at a position indicated by the identification number of each common gene in the second expression rank sequence data;
From the first expression rank sequence data and the second expression rank sequence data, the data of the expression rank at the corresponding position in each sequence data is read, and based on the read difference in the expression rank, A gene expression profile comparison program for obtaining the correlation by calculating a rank correlation coefficient with a question profile.

Images