While a biologist can perform a quick MedLine search to assess prior research in an area of interest, bioinformatics developers don’t have similar access to a centralized depository of software, according to Talapady Bhat, a researcher at the biotechnology division of the National Institute of Standards and Technology. As a result, Bhat said, a large number of software development efforts are unnecessarily duplicated in academic and nonprofit environments.
In response, Bhat and his colleagues at NIST are building a large-scale public repository called the Bioinformatics Software Resource (BISR). The database, available at bioinfo.nist.gov/BISR, will serve as a centralized archive for public domain software for bioinformatics and computational biology. BISR already offers 424 software packages, classified under seven categories: general tools, modeling and simulation, phylogenetics, sequence analysis, structural analysis and verification, structure determination, and visualization.
Bhat said the project aims to become for bioinformatics software what GenBank has become for genomic data. “Once software is deposited in BISR, it will be there for life,” said Bhat.
The NIST team has embarked upon a three-stage plan to reach this goal. The first stage is to direct users to available resources through web links. The majority of the packages currently available through BISR are available in this way. The second stage is to develop and test a web-based interface for archiving and distributing the resources from a centralized repository. The interface will allow authors to deposit source code, tools, documentation, and test data, and the database will hold and distribute the source code for others to access.
The third stage, which Bhat singled out as being particularly important, is to design and develop an application program interface for data exchange between different software. Bhat noted that the API would address one of bioinformatics’ more pervasive problems: the brief life expectancy of data formats, which requires constant updating of new features into application programs. If all BISR applications plug into a common API, as the developers propose, any evolutionary format changes could be made once within the API, leaving the application programs untouched.
Bhat said that a prototype of the API is currently available, but a good amount of work remains to be done.
The project has received a two-year $300,000 Advanced Technology Program grant as well as a grant from the Systems Integration for Manufacturing Applications program to support development of the BISR.The project team is currently seeking additional funding, but Bhat noted that what the BISR needs most right now is greater involvement from the bioinformatics community. “We need people to deposit new software packages and to develop application interfaces,” he said.
— BT
