Existing e-value correction software and their features

mpiBLAST [9], a parallel implementation of NCBI BLAST on a cluster, partitions the database and performs BLAST searches against these partitions in parallel. For accurate e-value correction, mpiBLAST requires prior knowledge of the entire database [9]. NOBLAST [19] corrects e-values when split databases are in use and results need to be aggregated. However, it does not work with Spouge statistics. We provide a detailed explanation of these tools in Section “Existing e-value correction software and their features” in the S1 File. Both tools (mpiBLAST and NOBLAST) provide exact e-value statistics for Karlin-Altschul statistics when knowledge about the entire database is available a priori. However, they are not useful when the database keeps changing or when two different search results against two different instances of similar databases need to be aggregated. Table 1 provides a high-level comparison between mpiBLAST, NOBLAST, and our iBLAST.

iBLAST supports e-value correction across time and space without requiring prior knowledge of the entire database while the other tools can perform e-value correction in limited scenarios.