8

I have some protein sequences and I want to build a position-specific scoring matrix (PSSM) for them and then upload this PSSM to NCBI PSI-BLAST. I used CHAPS program for this pupose but uploading the output PSSM gave me an error in NCBI PSI-BLAST. Do you know any tool or webserver for getting PSSMs for a group of protein sequences which then can work in NCBI PSI-BLAST? This is a part of my PSSM file:

PssmWithParameters ::= {
pssm {
    isProtein TRUE ,
    numRows 28 ,
    numColumns 131 ,
    byRow FALSE ,
    query
    seq {
        id {
            other {
                accession "WP_000208753" } } ,
        inst {
            repr raw ,
            mol aa ,
            length 131 ,
            seq-data
            ncbieaa "MTTKRKPYVRPMTSTWWKKLPFYRFYMLREGTAVPAVWFSIELIFGLFALKNGPEAW
AGFIDFLQNPVIVIINLITLAAALLHTKTWFELAPKAANIIVKDEKMGPEPIIKSLWAVTVVATIVILFVALYW" } } ,
    intermediateData {
        freqRatios {
            { 0, 10, 0 } ,
            { 564418841, 10, -10 } ,
            { 0, 10, 0 } ,
            { 11768571, 10, -9 } ,
            { 185838265, 10, -10 } ,
            { 31496547, 10, -9 } ,
            { 3872857, 10, -8 } ,
            { 291750464, 10, -10 } ,
            { 128450763, 10, -10 } ,
            { 759856221, 10, -10 } ,
            { 359173937, 10, -10 } ,
            { 179865517, 10, -9 } ,
            { 14537895, 10, -8 } ,
            { 212921456, 10, -10 } ,
            { 220554141, 10, -10 } ,
            { 368516324, 10, -10 } ,
            { 319343525, 10, -10 } ,
            { 426173953, 10, -10 } ,
            { 463659523, 10, -10 } ,
            { 817322186, 10, -10 } ,
            { 811693964, 10, -11 } ,
            { 0, 10, 0 } ,
            { 227810064, 10, -10 } ,
            { 0, 10, 0 } ,
            { 0, 10, 0 } ,
            { 0, 10, 0 } ,
            { 0, 10, 0 } ,
            { 0, 10, 0 } ,
            { 0, 10, 0 } ,
            { 768325726, 10, -10 } ,
            { 0, 10, 0 } ,
            { 1425562, 10, -8 } ,
            { 372685285, 10, -10 } ,
            { 466727421, 10, -10 } ,
            { 185741084, 10, -10 } ,
            { 427328646, 10, -10 } ,
            { 122594965, 10, -10 } , 
M__
  • 12,263
  • 5
  • 28
  • 47
Sara
  • 777
  • 1
  • 6
  • 18
  • Which error did it give you the NCBI PSI-BLAST? Could you paste here the image with the options used? What have you tried to overcome this problem? – llrs Jun 30 '17 at 14:15
  • The error was: Message ID#9 Error: PSSM processing error: Incomplete PSSM with Parameters provided on input: missing score matrix – Sara Jun 30 '17 at 14:25
  • So it seems that you need to provide the PSSM in certain format and CAPS didn't manage to create it in the expected format. Could you post the PSSM here (or in an accessible way) ? Now that I look in the help I couldn't find any reference to a specific format (just that it should be from their own output ...) – llrs Jun 30 '17 at 14:34
  • I do not exactly know how to post PSSM file here. Could you please guide me how to load a file here? – Sara Jun 30 '17 at 14:51
  • You can't but you can copy paste the content or add a link to dropbox or a similar server – llrs Jun 30 '17 at 14:58
  • @Llopis it should be fine posted here as code. – terdon Jun 30 '17 at 16:09
  • 1
    @Sara see formatting tools for help on formatting your posts. For things like this, you want to use code ({}) which uses a fixed-width font. – terdon Jun 30 '17 at 16:10
  • I think you could use standalone blast, one example here https://www.biostars.org/p/15482/ – friveroll Jul 02 '17 at 06:50

3 Answers3

2

If you take a look at my answer in this BioStars post, you can generate a PSSM using AlignIO in Biopython:

llrs
  • 4,693
  • 1
  • 18
  • 42
Joe Healey
  • 335
  • 2
  • 10
1

You can use biopython's code for generation of PSSM. Here is the link for the source code which can help you.

Sidra Younas
  • 503
  • 2
  • 13
1

I'm not 100% sure what you're trying to do with regards to your search in the first place, but I recommend using Sean Eddy's program Jackhammer it's similar to PSI-BLAST but uses Hammer HMM underpinnings.

As input it can take a single sequence, a multiple sequence alignment, or an existing HMM and then allow you to search in an iterative model building way. You can also download an HMM too at any stage in the iterative search, as well as the results / alignment.

Matt Bashton
  • 1,069
  • 6
  • 16