Changes between Initial Version and Version 1 of Ticket #7358, comment 9


Ignore:
Timestamp:
Aug 12, 2022, 4:25:32 PM (3 years ago)
Author:
Tom Goddard

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #7358, comment 9

    initial v1  
     1** BLAST test with !AlphaFold v3 database**
     2
    13I tested ChimeraX BlastProtein searching the 214 million sequence AlphaFold database.  Uniprot K0F9C8 with 451 amino acids took about 30 minutes and 64 Gbytes of memory on the server.  I updated the BlastProtein code to handle parsing the AlphaFold version 3 Fasta file title lines which have a different format than version 2.  The changes allow searching the version 3 or version 2 databases although there is currently no way to select the version, it is hard coded in blastprotein/job.py.  A blastprotein command argument could be added to allow searching either.  This code is ready to check in.  But I haven't checked it in because it seems like a step backwards to go from a K0F9C8 search that took 10 seconds in version 2 to taking 30 minutes (200 times longer) in version 3 to find the same structure.  Needs some more thought.