evolutionary amino acid position conservation

[BioSlayer]

I have E.coli protein sequences and want to compute the range of positional conservation of one amino acid (valine) across bacterial swissprot proteins. I don't really care what other amino-acids replace valine's position and therefore tools like SIFT, PROVEAN, PANTHER aren't useful as they expect 'interesting substitutions' and positions to be provided.

Eventually I used hmmer to searche for sequence homologs and generate alignments. I am at a bit of a loss how to proceed from here. Generating logos of the alignments is one option but to visually inspect them is time consuming.

What do you suggest ? Is there a tool that given a protein sequence and/or alignment can actually assess an amino-acid conservation for a large group of sequences?

[Jean]

It's mainly visual, but JalView should give you a conservation score for a position over a MSAS

[BioSlayer]

Thanks Jean but a visual inspection is a lengthy error prone eye-straining exercise so I thought of another strategy meanwhile. I started with constructing MSA alignments against a database of related species - uniprot bacterial knowledgebase - and then extracting that column in the alignment relating to my amino-acid of interest.

My approach to alignment is iterative-refinement in order to get as much a high quality alignment as possible. Jackhmmer is pretty neat because it is possible to iterate until no more alignment candidates from the database are found and tweak the inclusion/exclusion thresholds and therefore maintain a hold on accuracy .

From the alignment step I can then filter by which column position shows a weak conservation for valine and assess replacing valines in that colum with other residues in the consensus.

What I can do now is look into how replacing valine with any amino acid from the consensus will affect the protein function (deleterious/netural impact). I chose PROVEAN because it allows exploring different types of mutations. I had to download the NR database from NCBI and other pre-requisites such CD-hits and ncbi-blast++. PROVEAN calculates a delta-bit score of alignments before and after replacing the molecule. In effect it is re-aligning the sequences that I already aligned in Jackhmmer all over once again which I find repetitive but not necessarily time consuming.