Skip to content

Latest commit

 

History

History
59 lines (41 loc) · 1.16 KB

README.md

File metadata and controls

59 lines (41 loc) · 1.16 KB

Needleman and Wunsch on DPU

This repository implements two different alignment workload on UPMEM PiM.

The first one named dpu_alignment compares sets of sequences with all sequences in the same set being compared with all the other in the same set.

The second one named dpu_16s is an exemple of comparing a database of 16S RNA, all against all comparison. This can be use for more than just 16S RNA sequences.

Libraries needed

  • UPMEM SDK
  • filesystem and span std library (gcc vesrion >=10)
  • yaml-cpp

Build

make

Run application

For set comparison:

./dpu_alignment

For 16S RNA:

./dpu_16s

Alignment parameters can be changed in params.yaml and 16s.yaml Output scores and cigars in scores.txt and cigars.txt respectively.

Number of ranks used also in yaml files.

A test dataset is available for the dpu_16s application.

Dataset format

Set comparison fasta file form

> set {set_number} ...
{sequence}
> set {set_number} ...
{sequence}
.
.
.

All sequences with same number will be pair-aligned, same set number sequences must be contiguous !

16S comparison fasta file form:

> ...
{sequence}
> ...
{sequence}
.
.
.