Calling Ebi'S Qblast In A Standalone Program

Question

[Graslevy]

I am currently exploring the possiblilty of developing a phylogenetic tool to support an ongoing research. Coding with Servlets, JSP and Java; To overcome the perfomance hit (in bootstrapping), I am considering using web services for sequence alignment and phylogenetic trees. My understanding is that using QBLAST does not require SOAP/REST but I want to better understand the response from QBLAST.

On the ebi website (http://www.ebi.ac.uk/Tools/webservices/services/msa/clustalw2_rest) the run method specifies 'sequence' as a parameter. This should be the result of the multiple sequence alignment. My concern is the format required for the clustal service. Will this maatch the response from QBLAST? Do I need to modify the alignment from blast with Jukes before sending to clustal?

Secondly, I humbly request some advise on whether to use this approach or push forward with coding Needleman-Wunsch as well as the phylograms?

Thanks!

Answer 1

[Michael Schubert]

What exactly should your tool do? It's hard to give any recommendations without knowing that.

Concerning QBLAST: Look at the bottom of the page (or the URL) and find the year there: 1999, which is 12 years old. In order you access the NCBI BLAST programmatically, you better have a look at:

• the NCBI URLAPI, which you can use from simple GET and PUT requests
• the recently announced SOAP interface to NCBI BLAST (just got this off the BioPython mailing list)

For the phylogenetic tree, are you sure you want to build it with clustalw? It does create a tree, however, a very crude one. A better choice would be e.g. the PHYLIP package.

[Graslevy]

Thanks for your comment, yes I noticed the qblast page. The tool is a required deliverable to satisfy the computing department. Its strange that they want an equal computing component in the project. Would you advise using the web services approach?

[Graslevy]

btw, the research is a phylogenetic analysis of Entomophthorales...

[Michael Schubert]

You didn't really say what the tool should do ;) - just find related sequences and construct a tree?

[Graslevy]

Yes Sir. It will allow for MSA using sequences from file or BLAST search and then construct/display a phylogram. I am aware that this is unnecessary considering I can get things done elsewhere but the computing compnent is required as stated earlier.

[Michael Schubert]

Then I'd go with URLAPI (remote), PHYLIP (local), and whatever feels most comfortable to stitch them together. You might also want to have a look at BioJava/Perl/Python for nice I/O handling.