Opened 4 years ago

Last modified 3 years ago

#6346 assigned enhancement

Speed up AlphaFold database search possibly with mmseqs2

Reported by: Tom Goddard Owned by: Tom Goddard
Priority: moderate Milestone:
Component: Structure Prediction Version:
Keywords: Cc:
Blocked By: Blocking:
Notify when closed: Platform: all
Project: ChimeraX

Description (last modified by Tom Goddard)

Currently the "alphafold match" command currently takes about 12 seconds, using BLAT (not BLAST) on 7PUA chain DX on the AlphaFold database of 1 million sequences. It hangs ChimeraX while waiting. It takes about the same time for searching 73 sequences (all of 7PUA). EBI plans to eventually extend to 100 million sequences. In July of 2021 the database was introduced with 350K sequences and increased to 1 million in January 2022, so it has increased by 650K in 6 months. Search is probably 3 times slower now.

Investigate other search programs that might be faster, such as HMMER or MMSEQS2.

Ideally we'd want search to run under 10 seconds even as more sequences are added.

The "alphafold search" command using BLAST currently takes about 6 seconds on 7PUA chain DX.

Change History (3)

comment:1 by Tom Goddard, 4 years ago

Description: modified (diff)

comment:2 by Tom Goddard, 4 years ago

Owner: changed from Goddard to Tom Goddard

comment:3 by Tom Goddard, 3 years ago

Summary: Speed up AlphaFold searchSpeed up AlphaFold database search possibly with mmseqs2
Note: See TracTickets for help on using tickets.