Hi, Peter,

You are right, I thought there was already a datatype defined for PSSM, but it turned out there isn't one yet. I attached a draft tool configuration file for psiblast, it was based on the blastp configuration file, please revise as you like.

The options that are specific to psiblast are the following:
1. num_iterations
2. pseudocount
3. inclusion_ethresh

Other options should be the same as the blastp program. So a typical command line would be the same as blastp except adding above options if non-default values are used. I can see there are two optional output files, one is the PSSM and the other is ASCII PSSM. Besides using a typical query input, psiblast can take a PSSM file or a multiple sequence alignment file as the input.

Luobin

On Tue, Feb 5, 2013 at 4:53 AM, Peter Cock <p.j.a.cock@googlemail.com> wrote:
On Mon, Feb 4, 2013 at 7:32 PM, Peter Cock <p.j.a.cock@googlemail.com> wrote:
>
>
> On Monday, February 4, 2013, Luobin Yang wrote:
>>
>> Hi, Peter,
>>
>> Thanks for pointing them out!
>>
>> The phi-blast and psi-blast actually share the same binary, both use the
>> psiblast program in the NCBI BLAST+ tools. When the option -phi_pattern is
>> used, then it does phi-blast searching.
>>
>> Best,
>> Luobin
>>
>>
>
> Could you give a few command line examples of how you typically use
> psiblast,  focussing in the main options only?
>
> I would find that helpful when putting together a new wrapper - since
> including all the options right away is overwhelming.
>
> Thanks,
>
> Peter

I suspect we would also need to define another datatype for the PSSM
pattern files. That makes things more complicated.

Peter