JP4413017B2 - Clustering program and clustering apparatus - Google Patents

Description

  The present invention relates to a clustering program for classifying a plurality of sequences according to the degree of homology, and more particularly, to a clustering program and a clustering apparatus capable of speeding up a “single linkage clustering” algorithm. is there.

  For example, the process of classifying cDNA end sequences according to the degree of homology with known sequences is called “EST clustering”. As a part of the algorithm for realizing “EST clustering”, “single linkage” is generally used. The “clustering” algorithm is often used. This “single linkage clustering” algorithm can be realized by “SingleLinkageClusterer” (ftp://ftp.tigr.org/pub/software/singleLinkageClusterer), for example.

  As a typical program for classifying cDNA terminal sequences according to the degree of homology with known sequences, “BLASTCLUST (ftp://ftp.ncbi.nih.gov/blast/executables/LATEST-BLAST)” is available. Yes (see Non-Patent Document 1). This is a program that simultaneously performs a task of calculating the homology between a base sequence or amino acid sequence and a known sequence and a task of classifying based on the calculation result. Specifically, Non-Patent Document 1 describes the following contents. This “BLASTCLUST” uses “single linkage clustering” to perform grouping of amino acid sequences based on the homology calculation result by the “BLAST” algorithm. The program also takes as input a “FASTA” format file consisting of a plurality of protein sequences, each with a unique sequence identifier. The program then outputs an array identifier file organized as a cluster.

  In addition, the above "BLASTCLUST" can start processing from "BLAST". On the other hand, since a hit list (queryID, hitID) is saved in case of a crash, give the hit list as an input. Thus, the processing can be started from “single linkage clustering”.

NCBI News Fall / Winter 2000 "http://www.ncbi.nih.gov/Web/Newsltr/FallWinter2000/fall_winter2000.pdf"

  However, the above-mentioned “BLASTCLUST” is convenient in that the work of calculating the homology between the cDNA end sequence whose function is unknown and the known sequence and the work of classifying based on the calculation result are performed simultaneously. However, there is a problem that even if it is not flexible, that is, only the function of “single linkage clustering” is desired, such a function is not provided.

  Further, in “SingleLinkageClusterer” which is one means for realizing the “single linkage clustering” algorithm, for example, cluster members are created by checking the sequences in the hit list with each other. However, in this example, since it is necessary to perform a round-robin confirmation process between sequences in the hit list as described above, there is a problem that the time required for the classification process increases.

  Recently, a “BLAST” dedicated machine that performs parallel processing using a PC cluster is generally available. In addition, when such a dedicated machine is used, it is possible to increase the processing speed up to the hit list creation. However, since the “single linkage clustering” algorithm after “BLAST” has a process of checking all sequences in the hit list, it is impossible to parallelize the processes. There is a problem that it is very difficult to speed up the processing (clustering processing) related to the “linkage clustering” algorithm.

  The present invention has been made in view of the above, and in the case of classifying cDNA terminal sequences whose functions are unknown according to the degree of homology with known sequences, "BLAST" and "single linkage clustering" An object of the present invention is to provide a clustering program and a clustering apparatus that can separate and further significantly reduce the processing time required for “single linkage clustering”.

  In order to solve the above-described problems and achieve the object, a clustering program according to the present invention is a clustering program for classifying a plurality of classification target sequences based on homology, and is included in the classification target sequence, for example. A hit list described in a state where identifiers of a plurality of reference sequences (reference sequences) and identifiers of sequences having high homology with the reference sequences are associated with each other is read based on the hit list, Grouping the identifier of the reference sequence and the identifier of the sequence having high homology with the reference sequence, setting an individual cluster number for each group, and further sorting the identifier set with the cluster number, An array in which an individual array number is set for each identifier, and then the identifier, cluster number, and array number are associated with each other in the order of the array numbers. Sequence number order data generation step for generating order data, array number order data storage step for storing the sequence number order data in the first area, the array number order data is read, and the sequence number order data is clustered A cluster number order data generation step for generating cluster number order data in which the identifier, the cluster number, and the array number are associated with each other, and the cluster number order data is stored in the second area. The cluster number order data storage step, the array number order data and the cluster number order data are read out, and based on these data, all identifiers belonging to a plurality of cluster numbers sharing the same array number are the same. Process to be a cluster member and describe the identifiers that make up the cluster member A cluster member determination step of generating a cluster member list, characterized in that to execute a cluster member list storage step of storing the cluster member list in the third region, to the computer.

  The clustering program according to the present invention is a clustering program for classifying a plurality of cDNA end sequences having unknown functions based on the result of homology calculation with known sequences, respectively. A hit list described in a state where the identifier of each cDNA terminal sequence constituting the identifier and the identifier of the known sequence obtained as a result of the homology calculation are associated with each other is read out, and the cDNA terminal sequence is read based on the hit list And the identifiers of known sequences having high homology with the cDNA end sequences are grouped, individual cluster numbers are set for each group, and the identifiers set with the cluster numbers are further sorted. An individual sequence number is set for each, and then the identifier, cluster number, and sequence number in that sequence number order. A sequence number sequence data generation step for generating sequence number sequence data described in association with each other, a sequence number sequence data storage step for storing the sequence number sequence data in a first area, and reading the sequence number sequence data, The cluster number order data is generated by sorting the array number order data in the order of the cluster number, and generating the cluster number order data in which the identifier, the cluster number, and the array number are described in association with each other in the cluster number order. The cluster number order data storage step stored in the second area, the array number order data and the cluster number order data are read out, and belong to a plurality of cluster numbers sharing the same array number based on these data All identifiers are processed as the same cluster member, and the cluster member A cluster member determination step of generating a cluster member list that describes the identifiers that make up the over, characterized in that to execute a cluster member list storage step of storing the cluster member list in the third region, to the computer.

  Further, in the clustering program according to the present invention, in the cluster member determination step, all identifiers belonging to a specific cluster number are grouped together as the same cluster member based on the cluster number order data, and the sequence number order data As a result, if the identifier constituting the cluster member also belongs to another cluster number, the same cluster is applied to all identifiers belonging to the other cluster number based on the cluster number order data. As a member, on the other hand, when the newly added identifier does not belong to any remaining cluster number, the cluster member addition process is stopped, and then the series of processes until the stop, It is characterized by continuing until classification of all identifiers is completed.

  In the clustering program according to the present invention, in the cluster member determining step, when an identifier belonging to a specific cluster number does not belong to any other cluster number (no sharing), only an identifier belonging to the specific cluster number And forming a cluster member.

  In the clustering program according to the present invention, the task of calculating the homology between the cDNA terminal sequence whose function is unknown and the known sequence (“BLAST”), and the task of classifying based on the calculation result (clustering process) The configuration is realized in a separated form. In addition, the clustering program according to the present invention, for example, when classifying cDNA terminal sequences whose functions are unknown according to the degree of homology with known sequences, first sets an individual sequence number for each identifier, The cDNA end sequences and known sequences having high homology with the cDNA end sequences are grouped, and an individual cluster number is set for each group. One cluster member is formed by all the sequences belonging to a plurality of cluster numbers sharing the same sequence number. As a result, the processing time required for “single linkage clustering” is significantly greater than when using the “BLASTCLUST” program, which is a representative program, and when compared to the conventional method of generating cluster members in the entire sequence. As a result, the processing time required for “EST clustering” can be greatly reduced.

  Embodiments of a clustering program and a clustering apparatus according to the present invention will be described below in detail with reference to the drawings. Note that the present invention is not limited to the embodiments.

  FIG. 1 operates as a clustering apparatus for classifying cDNA terminal sequences with unknown functions according to the degree of homology with known sequences (corresponding to “EST clustering”), and can implement the clustering process of the present invention. It is a figure which shows the structure of a general computer system. In this example, as an example, the case where the cDNA terminal sequence is classified according to the degree of homology with the known sequence will be described. However, the sequence to be classified is not limited to the cDNA sequence. The processing in the examples may be applied to classification of base sequences other than cDNA and amino acid sequences. Further, the sequence to be classified does not necessarily need to be a sequence whose function is unknown.

  This computer system includes, for example, a control unit 101 including a CPU, a memory unit 102, a display unit 103, an input unit 104, a CD-ROM drive unit 105 (may be a DVD drive unit or an FD drive unit), a disk A unit 106 and an external I / F unit 107 are provided, and these units are connected to each other via a system bus A.

  In FIG. 1, the control unit 101 executes the clustering program of the present invention. The memory unit 102 includes various memories such as a RAM and a ROM, and stores programs to be executed by the control unit 101, necessary data obtained in the course of processing, and the like. The display unit 103 is composed of a CRT, LCD (liquid crystal display panel), or the like, and displays various screens to the user of the computer system. The input unit 104 includes a keyboard, a mouse, and the like, and is used by a computer system user to input various information. The illustrated CD-ROM 200 stores a program (clustering program) describing the clustering process of the present invention.

  Here, an operation example of the computer system until the clustering program of the present invention becomes executable will be described. First, in the computer system configured as described above, a clustering program is installed in the disk unit 106 from the CD-ROM 200 set in the CD-ROM drive unit 105. Then, the program read from the disk unit 106 is stored in the memory unit 102 when the computer system is activated (in the case of a dedicated machine) or when the program is executed. In this state, the control unit 101 (CPU) executes the clustering process of the present invention in accordance with the program stored in the memory unit 102.

  In the present invention, a program describing the clustering processing of the present invention is provided on the CD-ROM 200, but the recording medium of this program is not limited to this, and depends on the computer constituting the system. For example, other recording media such as a magnetic disk such as a floppy (registered trademark) disk, a magneto-optical disk, and a magnetic tape may be used. Moreover, it is good also as using the program provided by transmission media, such as an email and the internet.

  Next, an outline of the “EST clustering” process and the positioning of the clustering program of the present invention in the “EST clustering” process will be described. FIG. 2 is a flowchart showing a processing procedure of the “EST clustering”.

  First, the control unit 101 of the computer system generates a file describing a cDNA terminal sequence in “FASTA” format, that is, a query.fasta file, according to a user instruction, and stores the query.fasta file in a predetermined area of the memory unit 102. (Step S1).

  Next, in accordance with a user instruction, the control unit 101 is a file in which a known sequence already published by a public organization such as NCBI and a cDNA terminal sequence read from the query.fasta file are combined, that is, db A .fasta file is generated, and the db.fasta file is stored in a predetermined area of the memory unit 102 (step S2).

  Next, the control unit 101 performs “BLAST” with the query.fasta as the query side and the db.fasta as the database side (step S3). Then, the hit list obtained as a result of the above “BLAST” (the hit list is “a list, table, or data that can be used to know the combination of queryID and hitID, any format may be used)” is stored in memory. (Step S4) FIGS.3-1, 3-2, and 3-3 are diagrams illustrating an example of alignment obtained as a “BLAST” result. In this embodiment, “BLAST” is adopted as a homology search means. However, the present invention is not limited to this. For example, a known search method such as “dynamic programming method” or “FASTA method” may be used. Good.

  Finally, the control unit 101 executes a clustering program (including “single linkage clustering”) of the present invention, which will be described later, based on the hit list (step S5). Then, the resulting cluster member list is stored in a predetermined area of the memory unit 102 (step S6). Below, the process of step S5 and step S6 which are the characteristic processes of this invention is demonstrated in detail.

  Next, the clustering program of the present invention (corresponding to steps S5 and S6) will be described in detail with reference to the drawings. FIG. 4 is a flowchart showing processing by the clustering program of the present invention.

  Here, it is assumed that a hit list (queryID, hitID) has already been generated based on the result of “BLAST” that calculates the homology between the cDNA end sequence whose function is unknown and the known sequence. It is assumed that the list is already stored in a predetermined area of the memory unit 102. FIG. 5 is a diagram showing an example of the hit list. This hit list includes a query ID (corresponding to 000001_queryID_1 (refer to SEQ ID No. 1), 000002_queryID,...) And a homology calculation result (corresponding to the “BLAST” result) shown in FIG. Data describing the combination of hitID (corresponding to AB000096 (see SEQ ID No. 2), AK004551, AK004675, etc.) shown in FIG. However, the format of the hit list is not limited as long as it is a list, table, or data that shows the combination of queryID and hitID. In other words, the “BLAST” result itself may be the format converted for clustering processing. Also, here, as an example, a case where “BLAST” is executed using a database published by NCBI will be described.

  First, in the control unit 101, the hit list is read from the memory unit 102, and based on the hit list, only the queryID is described (see list_qid in FIG. 6) and only the hitID is described (see list_hitid in FIG. 7). ) And are stored individually in a predetermined area of the memory unit 102 (step S11).

  Next, the control unit 101 reads the hit list, list_qid, and list_hitid from the memory unit 102, and uses these pieces of information to identify an identifier (corresponding to the queryID and hitID) and an array (hereinafter simply referred to as an array). Is associated with the cluster number for classifying the above-mentioned cDNA terminal sequence and known sequence). That is, the cluster number “1” is assigned to the first queryID, and thereafter the cluster number is incremented every time the queryID is switched. Then, the associated data is stored in a predetermined area of the memory unit 102 (step S12). FIG. 8 is a diagram illustrating an example of data in a state where an identifier and a cluster number are associated with each other. Here, based on the “BLAST” result, a process of dividing sequences having high homology into the same group is performed. For example, the cluster number corresponding to 000001_queryID, AB000096, AK004551, AK004675 is set to “1”, and The cluster number corresponding to BC030169 is set to “2”, and thereafter, the cluster numbers corresponding to 000003_queryID and each hitID, 000004_queryID and each hitID, 000005_queryID and each hitID, etc. , “5”... (Increment). At this stage, all the sequence numbers are described as “−1”.

  Next, the control unit 101 reads data in a state where the identifier and the cluster number are associated from the memory unit 102, and sorts the data by the identifier. Then, the sorted data is stored in a predetermined area of the memory unit 102 (step S13). FIG. 9 is a diagram illustrating an example of the sorted data.

  Next, the control unit 101 reads the sorted data from the memory unit 102, sets an individual array number for each identifier in the array number column of the data, and associates the identifier with the array number. Here, array numbers corresponding to the respective identifiers are set to “0”, “1”, “2”,... In the order after the sorting, and the data after the array number setting is set as cluster_1 in a predetermined area of the memory unit 102. Store (step S14). At this time, if there is a record (row) having the same cluster number and array number, the record is deleted. Such duplication of the same record may be deleted from the hit list in advance. FIG. 10 is a diagram illustrating an example of cluster_1. The plurality of identical identifiers shown in the figure indicates that the identifiers are associated with a plurality of cluster numbers in an overlapping manner.

  Next, the control unit 101 reads the cluster_1 from the memory unit 102, sorts the cluster_1 by the cluster number, and stores the sorted data as cluster_2 in a predetermined area of the memory unit 102 (step S16). At this time, if there is a record (row) having the same cluster number and array number, the record is deleted. FIG. 11 is a diagram illustrating an example of cluster_2.

  Next, the control unit 101 executes “single linkage clustering” using the cluster_1 and cluster_2 (step S17). Here, the procedure of “single linkage clustering” of the present embodiment will be specifically described. FIG. 12 is a diagram illustrating a specific example of the “single linkage clustering” procedure of the present embodiment. FIG. 12 is a diagram for simply explaining the “single linkage clustering” of the present embodiment, and the correspondence between the illustrated array number and the cluster number is different from that in FIGS. 10 and 11 described above. .. Correspond to the cluster number of cluster_2, and the array numbers 0, 1, 2,... Correspond to the array number of cluster_1.

  For example, if the array element numbers corresponding to the identifiers (arrays) belonging to the cluster number “1” are “0”, “1”, “2”, these identifiers form the same cluster member. For example, if the identifier corresponding to the array number “2” also belongs to the cluster number “4”, all the identifiers (array numbers “2” and “6”) belonging to the cluster number “4” Are added as the same cluster members as above. Further, when an identifier belonging to the cluster number “4”, for example, an identifier corresponding to the array number “6” belongs to another cluster number, the identifier belonging to the cluster number is also added as the same cluster member. Thereafter, the cluster member addition process is continued until a newly added identifier does not belong to another cluster number. For example, referring to cluster_1, it can be immediately determined whether or not the identifier corresponding to the array number “6” belongs to a cluster number other than the cluster number “4”. It is easy to determine “a state that does not belong”.

  Further, when the array element numbers corresponding to the identifiers (arrays) belonging to the cluster number “2” are “3”, “4”, “5”, a cluster member different from the above is formed with these identifiers. . For example, if the identifier corresponding to the array number “5” also belongs to the cluster number “5”, all the identifiers (array numbers “5” and “8”) belonging to the cluster number “5” Are also added as the same cluster member. Further, when an identifier belonging to cluster number “5”, for example, an identifier corresponding to array number “8” belongs to another cluster number, the identifier belonging to that cluster number is also added as the same cluster member. Thereafter, the cluster member addition process is continued until a newly added identifier does not belong to another cluster number. When an identifier belonging to a specific cluster number does not belong to any other cluster number (no sharing), a cluster member is formed only with the identifier to which the specific cluster number belongs.

In this way, processing is performed for all cluster numbers, and the result is stored as cluster_3 in a predetermined area of the memory unit 102 (step S17). Thereby, the cDNA terminal sequence whose function is unknown can be attributed to the cluster member including the known sequence. FIG. 13 is a schematic diagram for explaining the contents of cluster_3. Actually, cluster_3 is expressed in the following list format.
cluster_3 [1] = {0,119978,126329,126352,3,119978,…}
cluster_3 [2] = {1,119979,132190,4,119979,…}

  Finally, the control unit 101 reads cluster_3 from the memory unit 102, generates a cluster member list based on the cluster_3, and stores the list in a predetermined area of the memory unit 102 (step S18). FIG. 14 is a diagram illustrating an example of the cluster member list.

  As described above, in this example, the work of calculating the homology between the cDNA terminal sequence whose function is unknown and the known sequence ("BLAST"), the work of classifying based on the calculation result (clustering process), It was set as the structure which implement | achieves in the form which isolate | separated.

  Further, in this example, when classifying cDNA terminal sequences whose functions are unknown according to the degree of homology with known sequences, first, an individual sequence number is set for each identifier (sequence), and The cDNA end sequences and known sequences having high homology with the cDNA end sequences are grouped, and an individual cluster number is set for each group. Then, one cluster member is formed by all identifiers belonging to a plurality of cluster numbers sharing the same sequence number. However, when an identifier belonging to a specific cluster number does not belong to any other cluster number (no sharing), a cluster member is formed only by the identifier belonging to the specific cluster number.

  As a result, compared to the case of using “BLASTCLUST”, which is a typical program, and the processing time required for “single linkage clustering”, compared to the conventional method of generating cluster members in the entire sequence. As a result, the processing time required for “EST clustering” can be greatly reduced.

  As described above, since the clustering program according to the present invention can group base sequences and amino acid sequences in a short time according to the degree of sequence homology, it is suitable for annotation of unknown function genes, DNA chip design, etc. Available to:

It is a figure which shows the structure of the general computer system which can implement | achieve the clustering process of this invention. It is a flowchart which shows the process sequence of "EST clustering". It is a figure which shows an example of the alignment obtained as a "BLAST" result. It is a figure which shows an example of the alignment obtained as a "BLAST" result. It is a figure which shows an example of the alignment obtained as a "BLAST" result. It is a flowchart which shows the process by the clustering program of this invention. It is a figure which shows an example of a hit list. It is a figure which shows an example of list_qid. It is a figure which shows an example of list_hitid. It is a figure which shows an example of the data of the state with which the identifier and the cluster number were linked | related. It is a figure which shows an example of the data after a sort. It is a figure which shows an example of cluster_1. It is a figure which shows an example of cluster_2. It is a figure which shows the specific example of the "single linkage clustering" procedure of a present Example. It is the schematic for demonstrating the content of cluster_3. It is a figure which shows an example of a cluster member list.

Explanation of symbols

101 Control Unit 102 Memory Unit 103 Display Unit 104 Input Unit 105 CD-ROM Drive Unit 106 Disk Unit 107 External I / F Unit 200 CD-ROM

Array number 1: 000001_queryID
Sequence number 2: AB000096

Claims (8)

A clustering program for classifying a plurality of target sequences based on homology,
Read a hit list described in a state where identifiers of a plurality of reference sequences (reference sequences) included in the classification target sequence and identifiers of sequences having high homology with the reference sequences are associated with each other, Based on the hit list, the identifier of the reference sequence and the identifier of the sequence having high homology with the reference sequence are grouped, and an individual cluster number is set for each group, and the cluster number is set. Sorts the identifiers, sets individual array numbers for each identifier, and then generates array number order data in which the identifiers, cluster numbers, and array numbers are described in association with each other in the order of the array numbers Steps,
A sequence number sequence data storage step for storing the sequence number sequence data in the first region;
Cluster number order data generation that reads the array number order data, sorts the array number order data in the order of the cluster number, and generates cluster number order data in which the identifier, the cluster number, and the array number are described in association with the cluster number order Steps,
A cluster number order data storage step for storing the cluster number order data in a second area;
The sequence number order data and the cluster number order data are read, and based on these data, all identifiers belonging to a plurality of cluster numbers sharing the same array number are processed as the same cluster member, A cluster member determining step for generating a cluster member list describing identifiers constituting the cluster member;
A cluster member list storing step of storing the cluster member list in a third area;
A clustering program for causing a computer to execute. A clustering program for classifying a plurality of cDNA end sequences having unknown functions based on the results of homology calculation with known sequences,
A hit list described in a state where the identifiers of the cDNA terminal sequences constituting the plurality of cDNA terminal sequences and the identifiers of known sequences obtained as the homology calculation results are associated with each other is read, Based on this, the identifier of the cDNA end sequence and the identifier of a known sequence having high homology with the cDNA end sequence are grouped, an individual cluster number is set for each group, and the cluster number is set. Array number order data generation step that sorts the identifiers, sets individual array numbers for each identifier, and then generates array number order data in which the identifiers, cluster numbers, and array numbers are described in association with each other in the order of the array numbers When,
A sequence number sequence data storage step for storing the sequence number sequence data in the first region;
Cluster number order data generation that reads the array number order data, sorts the array number order data in the order of the cluster number, and generates cluster number order data in which the identifier, the cluster number, and the array number are described in association with the cluster number order Steps,
A cluster number order data storage step for storing the cluster number order data in a second area;
The sequence number order data and the cluster number order data are read, and based on these data, all identifiers belonging to a plurality of cluster numbers sharing the same array number are processed as the same cluster member, A cluster member determining step for generating a cluster member list describing identifiers constituting the cluster member;
A cluster member list storing step of storing the cluster member list in a third area;
A clustering program for causing a computer to execute. In the cluster member determination step,
Based on the cluster number order data, all identifiers belonging to a specific cluster number are grouped as the same cluster member, and further, as a result of checking the sequence number order data, the identifier constituting the cluster member is another cluster number. Are added as the same cluster member for all the identifiers belonging to the other cluster numbers based on the cluster number order data, while the newly added identifiers are Stop the cluster member addition process when it does not belong to any cluster number,
3. The clustering program according to claim 1, wherein a series of processes until the stop is continued until classification of all identifiers is completed. In the cluster member determination step,
The cluster member is formed only by an identifier belonging to the specific cluster number when an identifier belonging to the specific cluster number does not belong to any other cluster number (no sharing). 3. The clustering program according to 3. In a clustering apparatus that classifies a plurality of target sequences based on homology,
Read a hit list described in a state where identifiers of a plurality of reference sequences (reference sequences) included in the classification target sequence and identifiers of sequences having high homology with the reference sequences are associated with each other, Based on the hit list, the identifier of the reference sequence and the identifier of the sequence having high homology with the reference sequence are grouped, and an individual cluster number is set for each group, and the cluster number is set. Sorts the identifiers, sets individual array numbers for each identifier, and then generates array number order data in which the identifiers, cluster numbers, and array numbers are described in association with each other in the order of the array numbers Means,
First storage means for storing the sequence number order data;
Cluster number order data generation that reads the array number order data, sorts the array number order data in the order of the cluster number, and generates cluster number order data in which the identifier, the cluster number, and the array number are described in association with the cluster number order Means,
Second storage means for storing the cluster number order data;
The sequence number order data and the cluster number order data are read, and based on these data, all identifiers belonging to a plurality of cluster numbers sharing the same array number are processed as the same cluster member, A cluster member determining means for generating a cluster member list describing identifiers constituting the cluster member;
Third storage means for storing the cluster member list;
A clustering apparatus comprising: In a clustering apparatus that classifies a plurality of cDNA end sequences with unknown functions based on the result of homology calculation with known sequences,
A hit list described in a state where the identifiers of the cDNA terminal sequences constituting the plurality of cDNA terminal sequences and the identifiers of known sequences obtained as the homology calculation results are associated with each other is read, Based on this, the identifier of the cDNA end sequence and the identifier of a known sequence having high homology with the cDNA end sequence are grouped, an individual cluster number is set for each group, and the cluster number is set. Array number order data generation means for sorting the identifiers, setting individual array numbers for each identifier, and then generating array number order data in which the identifiers, cluster numbers and array numbers are described in association with each other in the order of the array numbers When,
First storage means for storing the sequence number order data;
Cluster number order data generation that reads the array number order data, sorts the array number order data in the order of the cluster number, and generates cluster number order data in which the identifier, the cluster number, and the array number are described in association with the cluster number order Means,
Second storage means for storing the cluster number order data;
The sequence number order data and the cluster number order data are read, and based on these data, all identifiers belonging to a plurality of cluster numbers sharing the same array number are processed as the same cluster member, A cluster member determining means for generating a cluster member list describing identifiers constituting the cluster member;
Third storage means for storing the cluster member list;
A clustering apparatus comprising: The cluster member determining means includes
Based on the cluster number order data, all identifiers belonging to a specific cluster number are grouped as the same cluster member, and further, as a result of checking the sequence number order data, the identifier constituting the cluster member is another cluster number. Are added as the same cluster member for all the identifiers belonging to the other cluster numbers based on the cluster number order data, while the newly added identifiers are Stop the cluster member addition process when it does not belong to any cluster number,
7. The clustering apparatus according to claim 5, wherein a series of processes until the stop is continued until classification of all identifiers is completed. The cluster member determining means includes
The cluster member is formed only by the identifier belonging to the specific cluster number when the identifier belonging to the specific cluster number does not belong to any other cluster number (no sharing). 8. The clustering device according to 7.

Images