Blastx versus nr?

Question

[Buffo]

Hi everybody,

I want to perform a blastx versus nr locally for 2200 sequences, it is normal to take long long time? expected time to make it is about 800 hours!!

Somebody has had the same problem?

Any suggestions?

[GenoMax]

Would depend on length of the sequences and power of the computer infrastructure you have available locally. Still 800 h sounds a bit long.

I second suggestion for DIAMOND, but two critical things to keep in mind. You will need to create an index for nr with DIAMOND. You would also need a substantial amount of RAM for DIAMOND searches (if I recollect right ~60-80G).

[Buffo]

Thanks, I had heard about that, but I was looking for a suitable solution with blast, I will try diamond.

Answer 1

[Sej Modha]

I'd use DIAMOND instead of BLASTX as it's much faster than standard BLAST.

Answer 2

[Jake Warner]

I use Plast when Blast is too slow for me. It's also much faster especially if you decrease the seed-use ratio a bit.