Hi, Peter,

What difficulties did you see to include psi-blast and phi-blast when you were working on the other programs in the NCBI+ suite? I would be interested in creating a tool configuration file for each of these two.

Thanks, Luobin

On Mon, Feb 4, 2013 at 10:30 AM, Peter Cock p.j.a.cock@googlemail.comwrote:

On Mon, Feb 4, 2013 at 5:20 PM, Luobin Yang yangluob@isu.edu wrote:

...

Hi, Galaxy developers,

I do not see a web interface for psi-blast in NCBI BLAST+ in the tool

shed.

...

Does it have one?

Thanks, Luobin

If you are asking about the ncbi_blast_plus suite in the Tool Shed, this has most of the key NCBI BLAST+ standalone tools now, but not yet psiblast or rpsblast.

Peter

P.S. When you said web-interface, did you mean to the NCBI online server? As far as I know there are no Galaxy interfaces to that - just for the standalone NCBI command line tools.

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

On Mon, Feb 4, 2013 at 11:14 AM, Peter Cock p.j.a.cock@googlemail.comwrote:

On Mon, Feb 4, 2013 at 5:43 PM, Luobin Yang yangluob@isu.edu wrote:

...

Hi, Peter,

What difficulties did you see to include psi-blast and phi-blast when you were working on the other programs in the NCBI+ suite? I would be

interested

...

in creating a tool configuration file for each of these two.

Thanks, Luobin

Hi Luobin,

At first glance, the binaries for psiblast, rpsblast, and rpstblastn don't look too different for wrapping. We would need to define a third loc file for the RPS BLAST domain databases (like the existing loc files for nucleotide and protein databases).

However, I'm not immediately sure where phi-blast existing with the standalone NCBI BLAST+ tools. Do you know?

Peter

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

On Tue, Feb 5, 2013 at 4:28 PM, Luobin Yang yangluob@isu.edu wrote:

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

Thanks Luobin,

I've checked that into my development repository (on my tools branch): https://bitbucket.org/peterjc/galaxy-central/commits/f8f43f8494abdd228998d4e...

That will allow me to track changes etc as we work on the datatypes. Note I am not yet including ncbi_psiblast_wrapper.xml on the Galaxy Tool Shed.

Regards,

Peter