Home

balaBLAST is a free useful tool for the parallel execution of NCBI BLAST. It exploits I/O parallelism and database partitioning, while guaranteeing the same output reporting of a single-core NBCI BLAST. Performance results of balaBLAST shows clear when compared with others similar tools, particularly for very large sequence databases.

balaBLAST allows the execution of NCBI BLAST on several machines simultaneously adopting a non-intrusive approach. The NCBI BLAST program is exactly the same used when users access the NCBI webpage, either for remote executions or for local download. Furthermore, balaBLAST remains unchanged even when the NCBI BLAST is updated, as soon as the database format does not change.

balaBLAST is particularly powerful for I/O workload balancing – thus its name! It deals well with usual data skew situations, like concurrency (from different processes and users at each processing node) and heterogeneity (when distinct hardware may influence the performances).

Another important feature of balaBLAST is its capability to recover from partial execution. This keeps the robustness of the process in case of failures – software, hardware and external, like power supply.

Key words: I/O Parallelism, Robustness, Database Partitioning,
Non-instrusiveness.