Local Copy Of Scan Prosite Not Returning Good Results

Question

[Radu]

I've downloaded the packages from here, and I'm running the scan using ./ps_scan myProtein.seq but the results are very different from the results I get using the online tool. Not only are they different but the results are obviously wrong and if I use the -s flag nothing is returned. I have all the necessary files in the same directory and I'm running Perl v5.10.1 on Ubuntu Server 11.04. Any help would be much appreciated.

Here is the FASTA file I'm trying to pass through: http://pastebin.com/CYXwNWba

[Reyhaneh]

I don't know the reason for this error but I have always found contacting them really useful. You may try to ask them as well. http://prosite.expasy.org/contact

Answer 1

[Hamish]

Well I can replicate the problem...

Scanning the sequence (RefSeq:NP_001380.2) using ScanProSite or InterProScan (with only the ProfileScan and PatternScan applications) gives 6 matches against PS50853 (FN3) and 9 matches against PS50835 (IG_LIKE). Both of these are profile matches, so methods which just search the patterns return no hits.

Downloading and setting up 'ps_scan', and performing the search with common matches disabled (-s) gives:

$ perl ./ps_scan.pl -s ../test.seq 
open2: exec of pfscan -flxz -v - prosite.dat L=-1 failed at ./ps_scan.pl line 1899

Since pfscan is used to perform the scan against the ProSite profiles, and the profiles contain the expected matches, the failure of pfscan is the source of the problem. The matches you see when '-s' is not used are pattern matches which are handled by ps_scan.pl. Trying to run pfscan directly:

$ ./pfscan -flxz -v ../test.seq prosite.dat L=-1

Gives the expected matches, plus a few which will be filtered out by ps_scan.pl. So the problem is related to how ps_scan.pl runs pfscan. If I specify the path to pfscan explicitly on the command-line:

$ perl ./ps_scan.pl -s ../test.seq --pfscan ./pfscan

It now works, and I get the expected matches.

In the README there is a note covering this:

You may need to edit the ps_scan.pl to provide absolute paths to the directory
where you have installed the pfscan and psa2msa executables, unless you have
stored them in a directory in your PATH.

And looking in the code for ps_scan.pl there is this section:

# change this to the absolute path to the programs,
# unless they are located in a directory in your PATH
# or use option --pfscan and/or --psa2msa to give the full path.
my $PFSCAN  = 'pfscan';
my $PSA2MSA = 'psa2msa';