how to reproduce query result
hi
given a fasta file with multiple proteins,
A:H:L MNASRFLSALVFVLLAGESTAWYYNASSELMTYDEASAYCQRDYTHLVAIQNKEEINYLNSNLKHSPSYYWIGIRKVNNVWIWVGTGKPLTEEAQNWAPGEPNNKQRNEDCVEIYIQRTKDSGMWNDERCNKKKLALCYTASCTNASCSGHGECIETINSYTCKCHPGFLGPNCEQAVTCKPQEHPDYGSLNCSHPFGPFSYNSSCSFGCKRGYLPSSMETTVRCTSSGEWSAPAPACHVVECEALTHPAHGIRKCSSNPGSYPWNTTCTFDCVEGYRRVGAQNLQCTSSGIWDNETPSCKAVTCDAIPQPQNGFVSCSHSTAGELAFKSSCNFTCEQSFTLQGPAQVECSAQGQWTPQIPVCKAVQCEALSAPQQGNMKCLPSASGPFQNGSSCEFSCEEGFELKGSRRLQCGPRGEWDSKKPTCSAVKCDDVPRPQNGVMECAHATTGEFTYKSSCAFQCNEGFSLHGSAQLECTSQGKWTQEVPSCQVVQCPSLDVPGKMNMSCSGTAVFGTVCEFTCPDDWTLNGSAVLTCGATGRWSGMPPTCEAPVSPTRPLVVALSAAGTSLLTSSSLLYLLMRYFRKKAKKFVPASSCQSLQSFENYHVPSYNV:EVALQQSGAELVKPGASVKLSCAASGFTIKDAYMHWVKQKPEQGLEWIGRIDSGSSNTNYDPTFKGKATITADDSSNTAYLQMSSLTSEDTAVYYCARVGLSYWYAMDYWGQGTSVTVSS:DIVMTQSPSSLTVTTGEKVTMTCKSSQSLLNSGAQKNYLTWYQQKPGQSPKLLIYWASTRESGVPDRFTGSGSGTDFTLSISGVQAEDLAVYYCQNNYNYPLTFGAGTKLELK
- query using colabfold server
colabfold_batch test.fasta ./test_output_query --msa-only - use local colabfold search and colabfold batch
a) search
colabfold_search --mmseqs /nfsdata/home/~~~/workspace/ColabFold/mmseqs/bin/mmseqs /nfsdata/home/~~~/workspace/IntFold/plabdab/test.fasta ../../colabfold_db_gpu test_output_search --gpu 1 --db-load-mode 2b) run colabfold_batch on the results of a)colabfold_batch ./test_output_search ./test_output_batch --msa-only
does not get me the same results from 1). are there any instructions on how to do so?
the server query results in
. ./log.txt ./config.json ./cite.bibtex ./A_H_L_env ./A_H_L_env/out.tar.gz ./A_H_L_env/uniref.a3m ./A_H_L_env/pdb70.m8 ./A_H_L_env/bfd.mgnify30.metaeuk30.smag30.a3m ./A_H_L_env/msa.sh ./A_H_L_pairgreedy ./A_H_L_pairgreedy/out.tar.gz ./A_H_L_pairgreedy/pair.a3m ./A_H_L_pairgreedy/pair.sh ./A_H_L.a3m ./A_H_L_coverage.png ./A_H_L.pickle
and the local version leads to
. ./log.txt ./config.json ./cite.bibtex ./2.pickle ./2.a3m ./2_coverage.png ./1.pickle ./1.a3m ./1_coverage.png ./A_H_L.pickle ./A_H_L.a3m ./A_H_L_coverage.png
the local colab search leads to
. ./A_H_L.a3m ./1.a3m ./2.a3m
The GPU mode and server differ due to their prefiltering strategies, the server use the CPU version. So, to reproduce the results, please use the CPU version. GPU should be more sensitive.