Changes between Version 5 and Version 6 of Ticket #7358, comment 2


Ignore:
Timestamp:
Jul 29, 2022, 4:52:18 PM (3 years ago)
Author:
Tom Goddard

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #7358, comment 2

    v5 v6  
     1EBI sequence search on AlphaFold database 214 million structures took 10 minutes with FASTA, 13 minutes for BLAST for 139 residue sequences.  EBI does not offer mmseqs2 search.  Details follow.
     2
    13The EBI AlphaFold database suggests using their general sequence search page
    24
    … …  
    3941
    4042
    41 Should also try a UniRef90 search.  It probably won't be any faster, but AlphaFold DB mostly covers UniRef90 and possibly EBI has UniRef90 search more heavily optimized.
     43Tried same search on UniRef90 using FASTA.  It probably won't be any faster, but AlphaFold DB mostly covers UniRef90 and possibly EBI has UniRef90 search more heavily optimized.  Search was on 147 million sequences versus 214 million for AFDB.  Took 7.5 minutes.
     44
     45{{{
     46139 residues in 1 query   sequences
     4750157842458 residues in 147407377 library sequences
     48 Tcomplib [36.3.8h May, 2020] (32 proc in memory [15G])
     49 start: Fri Jul 29 23:36:13 2022 done: Fri Jul 29 23:43:42 2022
     50 Total Scan time: 1400.270 Total Display time:  0.050
     51
     52Function used was FASTA [36.3.8h May, 2020]
     53}}}
     54
     55using command
     56
     57{{{
     58cat fasta-E20220730-003038-0088-44777607-p1m.sequence | $APPBIN/fasta:36.3.8h /fasta36 -l $DATA_CURRENT/fastacfg/fasta3db -L -T 32 -p -s BL50 -f -10 -g -2 -E "10.0 -1.0" -F 0.0 -b 50 -d 50 -m "F9B fasta-E20220730-003038-0088-44777607-p1m.m9" -m "F10 fasta-E20220730-003038-0088-44777607-p1m.m10" -z 1 \@:1- +uniref90+ 2
     59}}}