Command line BLAST is very slow

0

Entering edit mode

2.6 years ago

eager_underdog • 0

Hi Fellow Bioinformaticians,

I'm running command line BLAST on an EC2 instance (r4.8xlarg, 32 cpus and 244 GB memory), and it is super slow. I have limited my search to humans only and used 8 threads. However, it looks like most of the processes are delayed as shown below. The command run here is blastp against the nr database.

enter image description here

I was wondering if you have any suggestions on accelerating the running time?

Thank you very much!

blastp BLAST blast blastn • 1.5k views

ADD COMMENT • link updated 2.6 years ago by GenoMax 147k • written 2.6 years ago by eager_underdog • 0

0

Entering edit mode

Please post your complete command line. What database are you blasting against and how much RAM is on this instance?

ADD REPLY • link 2.6 years ago by GenoMax 147k

0

Entering edit mode

Details added, thanks!

ADD REPLY • link 2.6 years ago by eager_underdog • 0

0

Entering edit mode

Are you using -num_threads option in your command line (which you did not include).

ADD REPLY • link 2.6 years ago by GenoMax 147k

0

Entering edit mode

Sure, num_threads is set to 8 as described in the quesiton.

ADD REPLY • link 2.6 years ago by eager_underdog • 0

0

Entering edit mode

Since you have 32 cores available you could have chosen more threads but that will require you to restart the job. It must be proceeding at the fastest pace it can based on your current selections.

ADD REPLY • link 2.6 years ago by GenoMax 147k

0

Entering edit mode

It would also be helpful to post the rest of (h)top screen. Are there any kswapd or kworker lines in it?

ADD REPLY • link 2.6 years ago by Mensur Dlakic ★ 28k

Login before adding your answer.