search

ebi-pf-team / interproscan

Genome-scale protein function classification
Apache License 2.0
300 stars 67 forks source link

Different results between local hmmer3 and interproscan #226

Closed JustineCharon closed 2 years ago

JustineCharon commented 3 years ago

Hi, I would need to run the equivalent of Interproscan on local protein profiles. When I run hmmscan (Hmmer3) with default parameters, I end up with many false positive, not find using InterProscan. Thus, I would be glad to know what are the parameters used in the hmmscan command to filter out false positive signals (I could not find this information in the Interproscan manual). Or if there is any way to run interproscan on local custom profile databases ?

Many thanks for your help

Justine

matthiasblum commented 3 years ago

Hi @JustineCharon,

InterProScan does not use the same hmmscan/hmmsearch options for all member databases.

For instance, hmmsearch is run with -Z 17929 --cut_ga when searching sequences against Pfam profiles, and with -Z 65245 -E 0.001 when searching sequences against CATH-Gene3D profiles.

Finally, most analyses include a post-processing step: not all domains reported by hmmscan are included in InterProScan's output.

You may try to use the -E 0.001 flag when running hmmsearch or hmmscan and see if it helps.