Finding variants for a particular gene/protein

Question

[Jenez]

Hello!

I was wondering if anyone had some input on the issue of finding variants that represent the total variation observed for a particular gene across species?

One idea that I have in mind is to utilize BLAST searches with certain interesting sequences as inputs. I would then attempt to filter this output based on some basic rules, for example e-value and maximal length difference. This I hope would yield a decent large set of all variants available in the database, on which clustering could be performed. Problem is, I do not know which database to utilize in this case, neither for protein sequences nor DNA sequences.

Are the refseq_ databases the way to go, or perhaps the non-redundant/nucleotide-collection database?

I have a vague idea on how to further restrict my searches based on taxonomy, but I'm not quite there yet seeing as which database to use is still unclear.

I would deeply appreciate your input!

Thank you!

Answer 1

[GenoMax]

"across species" is vague since for longer evolutionary distances you would need to do careful analysis but ..

At the protein level (which is where you would want to do this) -

NCBI has already done this for Eukaryotic species: http://www.ncbi.nlm.nih.gov/homologene (e.g. http://www.ncbi.nlm.nih.gov/homologene/135998)

Ensembl has something similar: ftp://ftp.ensembl.org/pub/release-82/emf/ensembl-compara/

Then there are the alignments available from UCSC for various genomes (human example here): http://hgdownload.soe.ucsc.edu/downloads.html#human

[Jenez]

I appreciate your input, but sadly it's bacterial sequences I'm interested in, which makes things a bit more tricky with horizontal gene transfer possible.

[GenoMax]

You can always do you own analysis but take a look at this as well: http://mbgd.genome.ad.jp/