How to accelerate the running process by adjust parameters

Question

How to accelerate the running process by adjust parameters

wk8910 opened this issue 6 years ago · 6 comments

Hi Heng,
I'm currently running minimap2 for very large genome. However, the running time is unbearable. I want to trade mapping quality with running time. Which parameters can be adjusted to get faster speed? Do you have any suggestions?
Thank you very much!
Best,
Kun

Answer 1 · 2019-02-01T10:00:50.000Z

The obvious options are -k and -w to control seeding. Increase those to be less sensitive, but faster.

Answer 2 · 2019-02-01T13:01:29.000Z

What's the size of your genome? What're the current parameters in use? How slow is "unbearable"?

Answer 3 · 2019-02-01T13:02:33.000Z

The obvious options are -k and -w to control seeding. Increase those to be less sensitive, but faster.

Thanks for your reply!

Answer 4 · 2019-02-01T13:04:23.000Z

What's the size of your genome? What're the current parameters in use? How slow is "unbearable"?

Hi Heng,
The genome size is about 40 Gigabase. I'm now trying to perform reads to reads alignment (The total amount of reads is about 2Tb). I'm currently using "-x ava-ont -t 8 -l 5000" and it need about 50 compute machine for more than one month.
Best,
Kun

Answer 5 · 2019-02-01T13:08:34.000Z

Get a machine with 256GB RAM and use -I50g. Increase the number of threads -t. Larger -k and -w will help as @armintoepfer said.

Answer 6 · 2019-02-01T13:09:49.000Z

Get a machine with 256GB RAM and use -I50g. Increase the number of threads -t. Larger -k and -w will help as @armintoepfer said.

Thanks very much!
I'm trying to test these new parameters.