Context
| Querying protein databases is a common bioinformatics task. Software are based on efficient indexing data structure associated to fast algorithms for computing pairwise sequence alignments. We have developed PANG (Protein Alignment with No Gap), a simplified version of the well-known BLAST software, to evaluate the performance of a PiM implementation for such application. The protein bank is split and indexed through the PiM memories. Queries are broadcast to all DPUs and only significant alignments are sent back to the host processor. Intensive protein sequence comparison achieved high speed-up compared to parallel Blast implementation. |  |
Code link
Talk
- D. Lavenier, Protein Alignment on UPMEM PiM Architecture, 1st minisymposium on applications and benefits of UPMEM commercial Massively Parallel Processing-in-Memory Platform, EURO-PAR 2023, Limassol, Cyprus, Aug. 2023 [slides]
