JP5453613B2 - Gene clustering apparatus and program - Google Patents

Description

  The present invention relates to a gene clustering apparatus and program for clustering a plurality of genes based on sequence similarity.

In order to estimate the function of a gene whose function is unknown, it is known that a technique of evaluating similarity to a known gene and performing clustering based on sequence similarity is effective.
Conventionally, a maximum saving method, a maximum likelihood method, a neighborhood connection method, and the like are used for gene clustering. These methods are common in that a phylogenetic tree is created while directly comparing the sequences of genes to be clustered. As an example using such clustering, a clustering and alignment program disclosed in Non-Patent Document 1 can be cited.

  In the conventional gene clustering method, a phylogenetic tree is created by paying attention to the base sequence of each gene, and estimating the time and context of mutation of each base sequence. However, with these methods, there is a problem that it is not possible to compare things that are significantly different from each other, such as those that are genetically separated or functions that are newly acquired after differentiation. . Conventional clustering is suitable for comparing gene changes that occur in an evolutionary process, that is, genes with relatively little change.

CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice; J.D.Thompson et.al .; Nucleic acids Research, 1994, Vol. 22, No.22 4673-4680.

  As described above, it is difficult to cluster genes that are evolutionarily separated by the method of clustering using all gene sequences as they are, as in the conventional clustering method. In addition, in order to narrow down gene functions and related proteins, it was very difficult to make a judgment based only on information contained in gene sequences.

  An object of the present invention is to provide a gene clustering apparatus and program that can discover genes having similar functions even in genes of evolutionary distant organisms.

  The second object of the present invention is to provide not only gene sequence information but also gene expression data to provide information that makes it easier to infer each function.

The gene clustering apparatus according to the present invention is a gene clustering apparatus that clusters a plurality of genes based on sequence similarity, and includes a motif search unit that searches for a motif sequence included in a gene sequence, and each gene sequence A motif score calculation unit for calculating a similarity score between two arbitrary genes by comparing motif sequences to be calculated, and an intergenic distance for calculating an intergenic distance between any two genes using the similarity score A calculation unit, a clustering processing unit that performs clustering of the plurality of genes based on the inter-gene distance, an expression data acquisition unit that acquires expression data of each gene from a gene expression data storage unit, and the acquired each Expression data display section that displays the expression data of each gene at the position corresponding to each gene Equipped with a.
In the present invention, the relationship between the gene expression data and the cluster analyzed for the gene similarity is compared using the motif included in the gene sequence as an index. Since genes that have similar functions even though they are evolutionarily separated often have similar motifs, the present invention can be used to discover functionally similar genes among a wide range of species and to estimate the functions of unknown genes. It is very effective. In addition, it is very effective to estimate the difference in the target protein and the like by finding the similar and similar functions at the motif level but different in the timing of gene expression.

  A gene clustering apparatus according to the present invention is a gene clustering apparatus that compares the results of clustering a plurality of genes using two or more feature vector quantities, and performs clustering using each of the feature vector quantities. Clustering processing unit to be executed, tone conversion unit for converting distance information of each sub-cluster into a one-dimensional tone number sequence based on the result of clustering, and clustering result using each feature vector amount And a parallel display unit for displaying in parallel the result of conversion into the one-dimensional gradation sequence.

  According to the present invention, it is possible to easily compare and grasp how similar two or more dendrograms created based on different data related to each gene are. In particular, it can be easily understood that the expression pattern differs depending on the expression time, expression site, and the like for gene groups that are known to be structurally similar from the dendrogram based on the motif. By using this information, we obtain important information such as differences in gene function, that is, the partner with which the generated protein interacts and the network in which it interacts. be able to.

A computer program according to the present invention is a program that causes a computer to function as a gene clustering apparatus that clusters a plurality of genes based on sequence similarity, and a motif search unit that searches for a motif sequence included in a gene sequence; A motif score calculator that calculates the similarity score of any two genes by comparing the motif sequences included in each gene sequence, and the intergenic distance between any two genes using the similarity score An intergene distance calculation unit that calculates a clustering processing unit that clusters the plurality of genes based on the intergene distance, and an expression data acquisition unit that acquires expression data of each gene from a gene expression data storage unit And the obtained expression data of each gene, To function as an expression data display unit for displaying the corresponding positions.
In the present invention, the relationship between the gene expression data and the cluster analyzed for the gene similarity is compared using the motif included in the gene sequence as an index. Since genes that have similar functions even though they are evolutionarily separated often have similar motifs, the present invention can be used to discover functionally similar genes among a wide range of species and to estimate the functions of unknown genes. It is very effective. In addition, it is very effective to estimate the difference in the target protein and the like by finding the similar and similar functions at the motif level but different in the timing of gene expression.

A computer program according to the present invention is a program that causes a computer to function as a gene clustering apparatus that compares the results of clustering a plurality of genes using two or more feature vector quantities, and each of the feature vectors. A clustering processing unit that performs clustering using a quantity, a tone conversion unit that converts distance information of each sub-cluster into a one-dimensional tone sequence based on the result of the clustering, and each of the feature vector quantities As for the clustering result using, the result converted into the one-dimensional gradation number sequence is made to function as a parallel display unit for displaying in parallel.
According to the present invention, it is possible to easily compare and grasp how similar two or more dendrograms created based on different data related to each gene are. In particular, it can be easily understood that the expression pattern differs depending on the expression time, expression site, and the like for gene groups that are known to be structurally similar from the dendrogram based on the motif. By using this information, we obtain important information such as differences in gene function, that is, the partner with which the generated protein interacts and the network in which it interacts. be able to.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
Embodiment 1 FIG.
FIG. 1 is a block diagram showing a functional configuration of a gene clustering apparatus 10 according to Embodiment 1 of the present invention. As shown in the figure, the gene clustering device 10 includes an input device 11, a user interface unit 12, a data access unit 13, a gene sequence storage unit 14, a score storage unit 15, a motif storage unit 16, a gene expression data storage unit 17, a motif A search unit 18, a motif score calculation unit 19, an intergene distance calculation unit 20, a clustering processing unit 21, an expression data acquisition unit 22, an output device 23, and an expression data display unit 24 are provided.

  The gene clustering apparatus 10 is obtained by causing a general-purpose personal computer to execute a predetermined program, for example. The user interface unit 12, data access unit 13, motif search unit 18, motif score calculation unit 19, intergene distance calculation unit 20, clustering processing unit 21, expression data acquisition unit 22, and expression data display unit 24 are computers according to programs. The modules of the operations performed by these processors are shown, and these actually constitute the processor of the gene clustering apparatus 10 as a whole.

The gene sequence storage unit 14, the score storage unit 15, the motif storage unit 16, and the expression data storage unit 17 are storage devices such as a hard disk of the gene clustering device 10.
The input device 11 is input means such as a keyboard, a mouse, and a touch panel, for example, and is used by the user to give processing instructions to the gene clustering device 10 and to input data and parameters. It is also possible to read data from a memory medium or the like via a USB (Universal Serial Bus) interface. The operation through the input device 11 by the user is controlled by the user interface unit 12.
The output device 23 is a display device, a printer, or the like.

  Next, the gene clustering process according to the present embodiment will be described. The gene sequence data to be analyzed and the gene expression data are stored in advance in the gene sequence storage unit 14 and the gene expression data storage unit 17 from the input device 11 via the user interface unit 12 and the data access unit 13. Is done. Score data for gene sequence comparison necessary for clustering is similarly input from the input device 11 and stored in the score storage unit 15. First, the sequence information of the gene group to be clustered is supplied from the gene sequence storage unit 14 to the motif search unit 18 via the data access unit 13.

  FIG. 2 is a diagram showing an example of gene groups to be clustered. Here, the gene number of the target gene and its species are shown. The example shown in FIG. 2 is a blast search (threshold 1e-30) for amino acid sequences of rice (Oryza Sativa), Arabidopsis thaliana, and red algae using the corn (Zea mays) ID1 (indeterminate1) gene as a query. ) And shows the hit genes.

Each gene sequence can be referred to, for example, at the following site.
Rice: http://rapdb.lab.nig.ac.jp/ (RAP1)
Arabidopsis: http://mips.gsf.de/proj/thal/db/ (MIPS)
Red algae: http://merolae.biol.su-tokyo.ac.jp/

The ID1 gene has been isolated as a gene that controls flowering in maize, and encodes a transcription factor having a zinc finger.
The method for selecting a gene group is not limited to the above method, and other sequence analysis methods may be used.

  Next, a motif search is performed in the motif search unit 18 for the supplied gene group. A motif is a sequence pattern corresponding to an active site or a functional region in a protein structure. The motif search can be performed using a technique such as MEME (Bailey and Elkan, 1994). FIG. 3 is a diagram illustrating an example of motif data obtained as a result of a motif search performed on the gene group partially shown in FIG. In the figure, the numbered squares correspond to individual motifs. For example, it is understood that the ID1 gene has a motif sequence represented by No. 5, No. 2, No. 3, No. 1, No. 7, No. 6, No. 18. In general, functionally similar genes often have the same motif even if they are genetically separated.

  By performing a motif search, it is possible to obtain information on partial sequences of various sizes, which are considered to contribute to determining the main structure / function, from the sequence of each gene. The obtained motif data is stored in the motif storage unit 16.

Next, the motif score calculation unit 19 compares all the genes to be clustered, and calculates a score representing the degree of similarity seen from the included motif sequences. For similarity score calculation, PAM (Point-Accepted Mutation, In Margaret O. Dayhoff, editor, Atlas of Protein Sequence and Structure, volume 5, pages 345-352. National Biochemical Research Foundation, Washington DC, 1978) and BLOSUM (Blocs Substitution Matrix, Henikoff and Henikoff (1992; PNAS 89: 10915-10919)) and the like can be used. The score storage unit 15 stores score data used in these methods.
In the present embodiment, score calculation is not performed for regions other than the motif. This means that the part other than the motif is regarded as score 0. Clustering is performed at high speed by narrowing down to the part where the sequence called motif is stored and calculating the score. If there is a further need, add not only the conserved sequence motif but also a function such as secondary structure prediction to extract the structural part that determines α-helix and β-sheet, and score them as motifs By giving, it is possible to perform not only functions but also structure-like clustering.

A similarity score calculation method will be described.
For example, the sequences of motif 1 included in gene 1 and motif 2 included in gene 2 are as follows.
Motif 1: WKCEKCAK
Motif 2: WKCDKCN

Since the first amino acid residue of motif 1 and motif 2 is W, referring to the column of W in the W row of the PAM40 matrix shown in FIG. It can be seen that the second amino acid residue is K for both sequences and the score is 6. When scores are sequentially obtained in this way and added, the scores of motif 1 and motif 2 are as follows.
Score = 13 + 6 + 9 + 3 + 6 + 9 + (− 3) = 43
In this way, a score is obtained for all the motifs included in gene 1 and gene 2 in a round-robin manner. Furthermore, the sum of the scores of all motifs is obtained and used as the similarity score between gene 1 and gene 2. Here, when calculating the optimal score in consideration of deletion and insertion of amino acid residues when comparing motifs, the algorithm Smith-Waterman method (Smith TF, Waterman MS (1981). “Identification of Common Molecular Subsequences”. Journal of Molecular Biology 147: 195-197.).

  FIG. 5 shows a part of the score matrix between genes determined as described above. FIG. 5 shows gene similarity scores for the four genes.

  Next, the intergene distance calculation unit 20 calculates the distance between each gene. The distance between genes can be defined in various ways. In the present invention, the Pearson correlation coefficient is used. This is a method of obtaining data of two arbitrary rows of the matrix shown in FIG. 5 and obtaining a correlation between elements. By using the correlation coefficient, the correlation becomes high for those having a relative motif similarity, and the correlation coefficient does not leave due to the bias of the absolute value. Even if there is a slight difference between a large number of common motifs and a small number of common motifs, the distance can be obtained while correcting with a common scale. In addition, a cosine coefficient can be used.

  Next, the clustering processing unit 21 performs clustering using a method such as the Ward method or the group average method using the distance value calculated by the intergene distance calculation unit 20. FIG. 6 shows a dendrogram of the clustering result. FIG. 6 suggests that the maize ID1 gene has a function similar to that of the Os10g0419200 gene. The Os10g0419200 gene encodes zinc finger protein, and the function of Os10g0419200 is given as Zinc finger and C2H2 type family protein, and it can be analogized that it actually has a similar function to ID1.

  Thus, according to the present invention, it is possible to perform a series of analyzes of extraction of motifs and clustering using the presence / absence and similarity of motifs as indices. A motif includes a conserved sequence pattern that is characteristic of a functional domain. By analyzing the motif as an index, genes that are functionally similar can be comparatively analyzed even if they are genetically separated. Analyzes using amino acid sequence substitution rates have existed so far, but methods for comparative analysis using the presence or similarity of motifs as indices have not been established, and functions that have been preserved between organisms in the future It can be used for gene analysis, function estimation of unknown function genes, and the like. Advances in DNA sequencing technology have led to the reading of genomes of a large number of species, and even if they are not necessarily genetically homologous, if functionally similar ones can be found by clustering, unknown genes It is very useful for analyzing the function of a sequence.

  It should be noted that the clustering method according to the present invention is not limited to gene motif information, but a numerical sequence pattern in which structural features are replaced with various index values such as α helix, β sheet, hydrophobic and hydrophilic areas. It can also be used as a target. Further, the gene sequence described in the present invention is a character string itself. Therefore, it is possible to replace the gene sequence as it is with clustering of character sequences. Needless to say, the present invention can be applied to any character information or numerical information sequence. For character strings, there is no problem even if the number of matched characters is used as a score, or a method of simply giving a constant score for each word existing in the dictionary. In the case of a numeric string, it is needless to say that the difference between the numerical values themselves or the square value thereof can be widely applied as a distance.

  Next, the expression data acquisition unit 22 extracts each expression data of these genes from the gene expression data storage unit 17. The gene expression data includes the gene expression level in pollen cells according to the maturation process, meiosis, tetramolecular, mononuclear, binuclear, and mature pollen, and the gene expression in tapetum. Data on the expression of metaphase, quadruple and mononuclear phase were used. The expression level is measured using a method using a DNA microarray, a method using RT-PCR, or a method such as reading the base sequence of the recovered mRNA with a DNA autosequencer and counting the number of mRNAs. The amount of mRNA expressed in the cell can be measured. FIG. 7 shows an example of measurement results using a DNA microarray. The expression data display unit 24 displays the expression data on the output device 23 together with the clustering result shown in FIG. Expression data is displayed at a position corresponding to each cluster. In the example shown in FIG. 7, each expression data is arranged in the horizontal direction of each leaf of the dendrogram. In some cases, two or three samples are used, and in that case, they are closely arranged. Furthermore, each expression level is displayed with a dark color corresponding to the measured expression level. Here, the darker the color, the greater the expression level. For example, in the early stage (71) of the pollen maturation stage of the plant, there are data for 4 samples, and in the tetramolecular period (72), the expression level is hardly changed and data for 3 samples is measured. In contrast, the latter half of the binuclear phase (73, 74) is shown in dark color, indicating that the expression level is increasing.

  FIG. 7 shows an example in which the group of genes from which the result that clustering is very similar is obtained increases at almost the same time in the expression level. FIG. 8 shows a case (81) in which the expression pattern is not conserved between genes that are paralogous (homologous sequences newly generated by gene duplication in a certain biological species). In addition, FIG. 9 shows an example (91) in which the onset time is slightly shifted due to paralogous material. Thus, by displaying the gene expression data side by side next to the dendrogram of the clustering result using the motif, the difference in gene behavior can be confirmed very easily.

  These drawing procedures are shown in FIG. First, in step 101, the expression data acquisition unit 22 acquires the expression data of each gene from the gene expression data storage unit 17. Next, in step 102, the expression pattern of the gene to which it belongs is displayed for each sub-cluster with reference to the created dendrogram structure. Furthermore, in step 103, the expression data display unit 24 arranges and draws the expression data beside the dendrogram. These results are the results of FIGS. Thereby, it is possible to visually compare the expression patterns of the genes in the subcluster.

  As described above, according to the present embodiment, the relationship between the gene expression data and the cluster in which the similarity of the genes is analyzed using the motif included in the gene sequence as an index is compared. Since genes that have similar functions even though they are evolutionarily separated often have similar motifs, the present invention can be used to discover functionally similar genes among a wide range of species and to estimate the functions of unknown genes. It is very effective. In addition, it is very effective to estimate the difference in the target protein and the like by finding the similar and similar functions at the motif level but different in the timing of gene expression.

  Moreover, by displaying together with the clustering result using gene motif information and the expression data, it is possible to consider by adding information on the actual movement of cells. Needless to say, the expression data can be combined according to the purpose of comparison, such as data acquired for each tissue, data acquired in time series, and data of different systems.

Embodiment 2. FIG.
In the second embodiment, clustering using gene expression data is performed in addition to clustering using gene motif information, and the results are displayed so that the results can be compared.
A method of comparing a plurality of clustering results will be described with reference to FIG. FIG. 11 shows an example in which the dendrogram calculated using the gene motif information by the clustering processing unit 21 is displayed at the top and the result of clustering using the gene expression data is displayed facing the bottom. In addition, heat map regions 115a, 116a, 115b, and 116b for comparing each cluster as described later are shown at the intermediate positions.

  Further, the left side 111 in the figure is a case where both clustering results are relatively similar. Also, the right side 112 in the figure is an example in which the clustering results are considerably different. First, 111 having a similar cluster structure will be described. The dendrogram 113a is the result of clustering using gene A, B, C, and D motif information. The distance between genes A and B is “3” as shown below the branch point of both branches. Furthermore, the distance from the center of gravity of genes A and B to C is “6”. Furthermore, the distance from the center of gravity of A, B, and C to D means “11”.

  In order to express the data of these distances in a compact manner, the expression data display unit (gradation conversion unit, parallel display unit) 24 corresponds to a color that becomes darker as the distance increases according to the gradation as shown in 115a. Colors under the gene. The relationship between each distance and gradation can be proportionally distributed so that the maximum distance of the target clustering result is assigned to “256” in the case of a 256 gradation display device. Further, if necessary, it is possible to perform correction so as to emphasize the short side and reduce the color difference on the far side by gamma correction. In this embodiment, since the distance “17” of the maximum value 114b is the maximum, the display gradation is calculated by correcting 255x (corresponding distance / maximum distance) so that the distance becomes 255.

  In the case of 111, when the dendrograms are similar as can be seen by comparing the heat map regions 115a and 116a, these heat map regions also have substantially the same gradation pattern. However, since the clustering result is different at 112, it can be easily identified that the tone patterns of the heat map regions 115b and 116b are different.

  Furthermore, although not shown numerically in the drawing of this embodiment, it is also possible to obtain the similarity of a plurality of clustering by obtaining the Pearson correlation coefficient of both using the numbers in this heat map area. For example, since 115a is (3, 3, 6, 11) and 116a is (5, 5, 7, 15), obtaining the Pearson correlation coefficient of both results in 0.9990. On the other hand, in 112, since 115b is (2, 2, 7, 11) and 116b is (17, 7, 7, 10), a negative correlation coefficient of -0.2768 is obtained as the Pearson correlation coefficient. When selecting a clustering result that is similar, for example, it can be easily selected by giving a threshold value such that the correlation coefficient is 0.7 or more. If it is desired to select one having a low correlation, it can be selected by selecting one having a low correlation or a value close to zero. By simply sorting from positive correlations to negative correlations so that they can be viewed in order from similar ones, the overall situation can be easily organized and grasped.

  In the figure, two dendrograms are compared. However, even if there are three or more dendrograms, it is possible to compare them with sufficient visual confirmation by simply arranging them side by side under the heat map regions 115a and 116a. is there. In addition, when numerically evaluating the degree of dispersion of three or more clustering results, the variance of gradations is obtained for each gene instead of Pearson's correlation coefficient, so that those with a large variation and those with no variation are sorted. Can be selected.

  For example, the average value of each gradation value is obtained as it is, and the cumulative value of the square of the deviation amount from the average value is averaged. As a result, it is possible to directly quantify the degree of variation in the heat map as viewed visually. As described above, according to the present invention, it is possible to compare and display the results of clustering with various pieces of information, and quickly find out what is similar to each other or what is not. This can be useful for discrimination when the proteins acting on the motif level are similar and possibly functionally similar, but the proteins that act are different.

  As described above, according to this embodiment, it is possible to easily compare and grasp how similar two or more dendrograms created based on different data related to each gene are. In particular, it can be easily understood that the expression pattern differs depending on the expression time, expression site, and the like for gene groups that are known to be structurally similar from the dendrogram based on the motif. By using this information, we obtain important information such as differences in gene function, that is, the partner with which the generated protein interacts and the network in which it interacts. be able to.

  In the present embodiment, clustering using similarity at the motif level and clustering results regarding gene expression levels are shown, but this can also be applied to various other feature quantities. For example, it goes without saying that the present invention is also used when various numerical groups (vector quantities) related to experimental conditions and vector quantities of the obtained experimental results are clustered and compared with each other.

FIG. 1 is a block diagram showing a functional configuration of a gene clustering apparatus according to Embodiment 1 of the present invention. FIG. 2 is a diagram showing an example of gene groups to be clustered. FIG. 3 is a diagram illustrating an example of a motif obtained by a search. FIG. 4 is a matrix table of PAM40. FIG. 5 is a diagram illustrating an example of a similarity score between genes. FIG. 6 is a dendrogram of gene clustering results. FIG. 7 is a diagram illustrating an example of a result of similarity of gene motif pattern and gene expression pattern. FIG. 8 is a diagram showing a case where an expression pattern is not conserved between paralogous genes. FIG. 9 is a diagram showing an example in which the onset time is slightly shifted due to paralogous things. FIG. 10 is a diagram showing a processing flow for performing expression pattern comparison processing in sub-clusters. FIG. 11 is a diagram illustrating an example in which a plurality of clustering results displayed in a heat map are displayed side by side.

Explanation of symbols

  10 gene clustering device, 11 input device, 12 user interface unit, 13 data access unit, 14 gene sequence storage unit, 15 score storage unit, 16 motif storage unit, 17 gene expression data storage unit, 18 motif search unit, 19 motif score Calculation unit, 20 Intergene distance calculation unit, 21 Clustering processing unit, 22 Expression data acquisition unit, 23 Output device, 24 Expression data display unit

Claims (6)

A gene clustering apparatus for clustering a plurality of genes based on sequence similarity,
A motif search unit that searches for a motif sequence that is included in a gene sequence and includes a sequence pattern corresponding to an active site or functional region in a protein structure;
A motif score calculation unit for calculating a similarity score between any two genes based on a motif sequence included in each gene sequence to be clustered ;
An intergenic distance calculation unit for calculating an intergenic distance between any two genes using the similarity score;
A clustering processing unit for clustering the plurality of genes based on the intergenic distance;
An expression data acquisition unit for acquiring the expression data of each gene from the gene expression data storage unit;
An expression data display unit for displaying the acquired expression data of each gene at a position corresponding to each gene;
The motif score calculation unit calculates a similarity score between all motif sequences included in any two genes, assumes that the similarity score of a region other than the motif sequence is zero, and calculates all the calculated motif sequences A gene clustering apparatus characterized in that a sum of similarity scores is used as a similarity score of the two genes . The clustering processing unit further performs clustering using a feature vector amount other than the inter-gene distance,
A gradation converter that converts the distance information of each sub-cluster into a one-dimensional gradation sequence based on the result of each clustering;
The gene clustering apparatus according to claim 1, further comprising: a parallel display unit that displays in parallel the results of conversion into the one-dimensional gradation sequence for each clustering result. The expression data display unit displays a plurality of expression data respectively obtained from a plurality of samples for each cell maturation process, along with the cell maturation process, and the expression data at the same maturation stage are displayed in close contact with each other. The gene clustering apparatus according to claim 1 or 2, wherein each expression data is colored with a concentration corresponding to the measured expression level. Computer
A program that functions as a gene clustering device that clusters a plurality of genes based on sequence similarity,
A motif search unit that searches for a motif sequence that is included in a gene sequence and includes a sequence pattern corresponding to an active site or functional region in a protein structure;
A motif score calculation unit for calculating a similarity score between any two genes based on a motif sequence included in each gene sequence to be clustered ;
An intergenic distance calculation unit for calculating an intergenic distance between any two genes using the similarity score;
A clustering processing unit for clustering the plurality of genes based on the intergenic distance;
An expression data acquisition unit for acquiring the expression data of each gene from the gene expression data storage unit;
The acquired expression data of each gene is made to function as an expression data display unit that displays it at a position corresponding to each gene ,
The motif score calculation unit calculates a similarity score between all motif sequences included in any two genes, assumes that the similarity score of a region other than the motif sequence is zero, and calculates all the calculated motif sequences A program characterized in that a sum of similarity scores is used as a similarity score of the two genes . The clustering processing unit further performs clustering using a feature vector amount other than the inter-gene distance,
The computer,
A gradation converter that converts the distance information of each sub-cluster into a one-dimensional gradation sequence based on the result of each clustering;
5. The program according to claim 4 , wherein each of the clustering results is caused to function as a parallel display unit that displays the result of conversion into the one-dimensional gradation number sequence in parallel. The expression data display unit displays a plurality of expression data respectively obtained from a plurality of samples for each cell maturation process, along with the cell maturation process, and the expression data at the same maturation stage are displayed in close contact with each other. 6. The program according to claim 4 or 5, wherein each expression data is colored with a concentration corresponding to the measured expression level.