Parallel blast

[necrolyte2]

No description provided.

[necrolyte2]

Waiting for VDBWRAIR/bio_bits#46 to be completed and also next version to be released

[necrolyte2]

I'm integrating changes from VDBWRAIR/bio_bits#61 just to ensure parallel_blast is actually finally correct

[necrolyte2]

TODO

• Run pbs job to ensure it works
• Compare results both from the diamond/blast results as well as the time taken

[necrolyte2]

I'm running a test script that will run par_block_blast.pl and parallel_blast with the same parameters using blastx on a fasta file that has 304 entries.

I'm running it on my local computer with 4 instances and have a job queued to run on the cluster with 10 cpus

Hopefully this will show for sure if parallel_blast is faster or not(especially now that we are using --round-robin)

[necrolyte2]

https://gist.github.com/necrolyte2/6ff2e4dab8183fd2a027

On my machine par_block and parallel_blast took almost the same amount of time.

• parallel_blast 4 cpu - 06:08:00
• par_block_blast.pl - 06:34:00

Running on our cluster with a single node and 10 cpu now and then will run parallel-blast with a few nodes to see how fast it speeds up

[necrolyte2]

Tracking progress in this badly formatted google doc
https://docs.google.com/spreadsheets/d/1D3VdTsJqBZ8zVpdkmj3VxWIqrX-UFCBTzehp5K2-FW0/edit?usp=sharing