JP2001167072A - Genetic data analytic system using network - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a genetic data analytic system using network, with which an operable minimum unit can be made small by changing a method, for paralleling computers and performing genetic information processing, the introduction cost can be extremely reduced and a computer system can be easily extended as well. SOLUTION: When there is a vacancy in the job of a computing server 2 itself, the computing server utilizes a network by itself and performs a job request to a server 1 for task managing. The server 1 for task managing changes the state of a task, to which the job is dispatched, and waits the message of computation end from the computing server 2. During this wait, the computing server 2 performs genetic information retrieval while utilizing the data base of a data base server or singly executes computing processing without utilizing the data base of the data base server 3. After the end of computation, the computing server 2 transmits computed result data to the sever 1 for task managing together with information on the success or failure of computation.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system for distributing and supplying processing tasks to a computer for gene analysis, and managing the processing steps together with the calculation results.

[0002]

2. Description of the Related Art Genetic information processing uses an enormous database and performs computations using complex algorithms.
At present, the size of gene databases is increasing year by year, despite the fact that the amount of processing per computer per day is limited. Therefore, in order to construct a genetic information processing system, it is necessary to predict the amount of processing that is increasing year by year and to design the system by predicting the amount of information several years later. At the present time, in order to process vast amounts of genetic data at high speed, at the present time, dedicated computers equipped with computer parallelization and special hardware, or
I use an ultra-high-speed computer.

[0003]

However, in order to introduce a system that will operate several years later, as in the prior art described above, it is necessary to initially introduce a very large system. For example, assuming that the system will be built three years from now, the system introduction cost and operation cost are irrelevant to the utilization rate even though only 1/3 of the functions are used in the first year It is calculated to pay three years in advance. Of course, even if the plan is changed in the middle, the cost of changing the system will be in proportion to the first system introduced.

[0004] Next, in order to process such enormous information at high speed, parallelization of computers and special hardware such as arranging a plurality of identical models, dividing one job and performing parallel processing are performed. Uses a special-purpose computer equipped with hardware or an ultra-high-speed computer. However, the construction of such a computer system requires enormous costs and makes maintenance and expansion difficult.

[0005] Further, since the minimum unit of parallelization has a large scale, initial introduction at low cost is impossible. Furthermore, there is a method of using special hardware, but if the hardware is changed, it is necessary to change the program and algorithm according to the maker, and the initial installation cost is very large.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problem, and the minimum unit is changed by changing the method of parallelizing computers, managing data, and executing genetic information processing. To provide a genetic data analysis system using a network that can be reduced in size, can easily expand and change the computer system without considering the type of computer, etc., and can greatly reduce the introduction cost. Aim.

[0007]

In order to achieve the above object, a gene data analysis system using a network according to the present invention comprises a gene data analysis system in which a task management server and a calculation server are connected by a network. It is characterized in that the calculation server processes the information processing on a job-by-job basis, and that the task management server allocates a job to the calculation server in response to a job request from the calculation server.

That is, in the conventional parallel processing, one job is divided, a divided job is allocated to a plurality of computers, the divided processing is performed, and the processing results output from each computer are collectively answered as one job. If the types, performance, and accuracy of these multiple computers do not exactly match each other, reliable processing results will not be obtained. However, in the present invention,
In the case of genetic information processing, the feature of having a large number of jobs is used, and calculation is performed by dividing into multiple spare calculation servers on a job basis. Even if they are different, they can be easily added later as a computer, and the degree of freedom of combination increases.
In this way, the time required for each job is not different from that before the conventional parallel processing, but the total time is faster due to the large number of jobs.

In order to use this method, it is necessary to manage job tasks and manage processing results. Therefore, job distribution is performed according to a request from the calculation server, and where the answer of the distributed job is stored is managed by the management server. Furthermore, since a general computer network is used, if it is connected by a network, it can be used regardless of the installation location.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a genetic data analysis system using a network according to the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing a system configuration of the present invention, and FIG. 2 is an operation flowchart of the present invention. In the present invention,
It utilizes two new technologies. One is a change in the parallel processing method, and the other is in the data management method. As shown in FIG. 1, the system includes a task management server 1 for supplying tasks, and a database containing a gene database commonly used by one or more network-connected computation servers 2 for performing a plurality of gene analysis processes. It comprises a server 3 and a client PC 4. However, the database server 3 can be shared by the calculation server 2. When the amount of calculation is small and the amount of data to be managed is small, the task management server 1, the calculation server 2, and the database server 3 can be the same computer.

The operation principle will be described below. Since a plurality of jobs are all managed by the task management server, all the tasks are stored in the RDB (Relationa
l Data Base) or register and manage in an independently constructed database (Fig. 2 (1), (2)). The method of this registration is to create a worksheet in advance according to whether to register from a file to a database as shown in FIG. 4 or to cut and paste directly into an input box and register into a database as shown in FIG. When data is sent to the task management server from an experimental device or a disk, the data is registered in the task management tables using a program that automatically registers the data in the database. When there is free space in its own job, the calculation server 2 makes a job request to the task management server 1 using the network itself (FIG. 2 (3)). The task management server 1 passes the job to the calculation server 2 (FIG. 2 (4)), changes the state of the task managed by the task management server 1, and waits for a calculation completion message from the calculation server 2. . In the meantime, the calculation server 2 performs a genetic information search using a database of the database server, or performs a calculation process independently without using a database of the database server 3. After the calculation is completed, the calculation server 2 transmits calculation result data to the task management server 1 together with information on the success or failure of the calculation (FIG. 2 (5)). And
For example, the result table 13 described below in the task management server is updated (FIG. 2 (6)).

As described above, instead of dividing one job into several jobs and assigning them to a plurality of calculation servers,
Originally, the characteristics of genetic information analysis, which has a large number of jobs, are used and assigned to multiple calculation servers on a job-by-job basis. Can be added. Conventionally, the allocation method also uses the management server to monitor the availability of the calculation server and assigns a job to an available calculation server. When the processing is completed, the calculation server transmits the result to the management server, and obtains the next job from the management server using the network.

The network protocol and method are not limited. For example, if the amount of processing is small or you want to use the Internet and perform processing from a remote location, you can use the http protocol and a Web robot.If the amount of processing is relatively large, use a TCP / IP socket. When high-speed processing is required, use the RDBMS (Relational Data Management System).
m), using TCP / IP, multi-protocol, etc.
To be able to communicate directly between databases,
It can be used according to the scale and environment. Depending on the capacity of the calculation server, it has the ability to process a plurality of jobs at a time, so if there is a function to provide a plurality of jobs to the management server at one time according to the capacity, the efficiency will be high.

If such a method is used, the task management server 1 does not need to manage the total number of each of the calculation servers and the availability of each of them. In addition to the selection, if you want to add a calculation server, you can add it as needed according to the processing size. Furthermore, if the job request program and the calculation processing program operate on the calculation server, the cost can be significantly reduced because the model of the calculation server does not matter. Further, if a calculation server can be connected to the network, the calculation server can be used even in a remote place, and the calculation server can be added and connected in a wide range. Next, in order to perform the above-described data management, the task management server 1, for example, as shown in FIG.
It is necessary to create a database structure as shown in. The calculation condition table 11 is a specific example actually used at the present time, which is a condition for homology search and alignment, but can be applied to other calculation methods.

The data table 12 mainly includes sequences of genes and amino acids used for calculation. The result table 13 uses data in the data table 12,
The data in various forms of the result processed under the specified calculation condition of the calculation condition table 11. In general,
In many cases, a plurality of the result tables 13 exist depending on the classification. Normally, data management is possible with the tables 11, 12, and 13, but a task list table 14 for sending a job to the calculation server is newly added. Of course, depending on the situation, tables 11, 12, 13
Is not created, and 11, 11,
Information 12 and 13 may be added.

Information is previously registered in these tables by some method, and when requested by the calculation server, data waiting for calculation is selected from the task list table 14 and a job is sent to the calculation server 2. The calculation status flag is updated along with the calculation server information (which computer is performing the processing). FIG. 6 shows an example of the task list table 14. In FIG. 6, the date when the task was registered, the name of the task, the ID, and the like are registered. The status is shown. Where SUCCESS
Indicates that the processing has been completed, and WAIT indicates a state of waiting for processing. Conventionally, the calculation result is usually managed by the calculation server 2, but in the present invention, the calculation result or a part of the information is stored in the result table 13 of the task management server 1.
To save.

[0017]

According to the genetic data analysis system using the network of the present invention, the minimum unit can be reduced by changing the method of parallelizing computers, managing data, and executing genetic information processing. In addition to this, the computer system can be easily expanded and changed without considering the type of computer and the like, and the introduction cost can be extremely reduced. Further, if a calculation server can be connected to the network, the calculation server can be used even in a remote place, and the calculation server can be added and connected in a wide range.

[Brief description of the drawings]

FIG. 1 is a system configuration diagram of a gene data analysis system using a network according to the present invention.

FIG. 2 is an operation flowchart of the present invention.

FIG. 3 is a database structure diagram of a task management server used in the genetic data analysis system using the network of the present invention.

FIG. 4 is a diagram showing a state in which data to be analyzed (retrieved) is selected by selecting a file in a database of a task management server and data is registered;

FIG. 5 is a diagram showing a state in which data to be analyzed (retrieved) is pasted into a database of a task management server and data registration is performed.

FIG. 6 is a diagram showing a state of a task list table.

Claims (1)

[Claims] In a genetic data analysis system in which a task management server and a calculation server are connected by a network, means for causing the calculation server to perform genetic information processing on a job basis, and task management by a job request from the calculation server. A genetic data analysis system using a network, wherein a server allocates a job to a calculation server.