On Jun 4, 9:48 am, "T Joshi" <tejalonl... from gmail.com> wrote:
> Anyways, my goal is to find out whether the motifs predicted by MEME
> (and other algorithms) from 5' UTR sequences are conserved across
> species. For this I have two sets of sequences:
> 1) 5' UTR sequences of homo sapiens (collected from BioMart) with
> Ensembl Gene ID.
> 2) I have downloaded sequences from the conservation track of UCSC
> extracted from 5' UTR (exon) region with the intersection with ensembl
>> I am applying motif discovery algorithms on the first set of
> sequences. Later, I do a multiple alignment of these motifs with the
> second set of sequences(conserved 5' UTR) using clustalw.
>> Is this approach a proper one?
Not really. Motifs as such are represented by position weight matrices
(save for few programs using graph representation). So instead of
clustalw you should be using i.e MAST program for MEME motifs.
Also I did not get if you are trying to do some kind of exhaustive
search for all motifs in all human 5'UTRs or you are more lucky and
got some subset of say co-expressed genes/ orthologues.
You may take a look at this wiki page (and several links from there)
just to get some more ideas:
may answer your questions.