SC17 Denver, CO

P18: A Parallel Python Implementation of BLAST+ (PPIB) for Characterization of Complex Microbial Consortia

Authors: Amina Jackson (Naval Research Laboratory), William Connor Horne (Naval Research Laboratory), Daniel Beall (Naval Research Laboratory), Kenneth Jiang (Naval Research Laboratory), William Judson Hervey (Naval Research Laboratory)

Abstract: Technological advancements in analytical instrumentation have enabled large-scale data acquisitions among the ‘-omics’ sciences of genomics, transcriptomics, and proteomics. An essential application among ‘-omics’ disciplines is the Basic Local Alignment Search Tool (BLAST) for functional inference of biomolecules. Though implementations of BLAST+ have been modified to address data volume growth, such improvements have neither been consistently maintained for high performance computing (HPC), nor have they been applied to complex microbiomes. Further, such implementations do not scale well to microbiomes of Naval interest on HPC systems in our hands.

Here, we compare 2 HPC implementations: BLAST+ and a Parallel Python Implementation of BLAST+ (PPIB) for protein functional inference. PPIB enabled functional inference of 2 complex microbiomes, which may be attributed to a combination of MPI and Python multiprocessing to query up to 3,600 proteins simultaneously. In contrast, BLAST+ did not complete functional assignments relative to PPIB at a comparable walltime.

Award: Best Poster Finalist (BP): no

Poster: pdf
Two-page extended abstract: pdf

Poster Index