Multiple Sequence Alignment tools

Question

[Rory Osborne]

Does anyone happen to know of any software/ online tool which allows me to do the following:

I would like to align several protein sequences (over 100) from one organism with a similar number from a different organism, to find which protein(s) from organism X are most like those from organism Y, or even which domains. I am already aware of the many MSA tools available online, but none of them exclude intra-species homology (I am only interested in where similarities occur between the two organisms). Checking protein by protein is obviously doable but would require an exhaustible number of alignments to be made. Ideally I would give this tool all of the protein sequences (FASTA) from one organism, and 1 sequence from the second organism, and it would return to me a list of proteins with the highest scoring homology, and where in the sequence the similarities occur. Alternatively I would give it all from both organisms and it would do a similar, although I appreciate this is far less likely to work.

Any suggestions would be greatly appreciated,

Thanks

[george.ry]

Unless I'm missing some complexity here, you don't need an MSA to achieve this. A local blast library will do this far more simply.

[Benn]

I agree with @george.ry, first use blast to find the most similar sequences. Then you can use these 'hit' sequences for a multiple alignment with your query sequence to compare it "better".

[Persistent LABS]

Right said george.ry and b.nota. Another way to look into this problem could be look for a Bidirectional Best Hits i.e. searching a sequence from orgY against orgX, if a hit is found, then take the same hit and search against orgY. If birectional hits are found, then it gives you better confidence of homology. The concept is similar to COG. Probably What Is The Best Method To Find Orthologous Genes Of A Species? can give even better information.

Thanks

Priyabrata

Persistent LABS

[Medhat]

You can use mafft then process the result to get what you need

Answer 1

[5heikki]

You could concatenate the proteins into one file and then cluster (e.g. with cd-hit) said file with high similarity percentage. Just be sure that you can tell from sequence names from which proteome the cluster members come from. Then you can examine the cluster output file..