Hi all,
I am looking for a piece of advice related to a efficient and accurate way of obtaining a multiple sequence alignment of all the 3'UTRs of a genome. I'm aiming to detect conserved and repetitive patterns that could be targets for regulatory elements, but I am encountering memory limitations (classic multiple alignment tools -such as Muscle- cannot handle files that contain more than 500 sequences or their size is over 10 Mb).
I would hugely appreciate any ideas or feedback.
Thank you very much in advance!
•
written
by
aramis.1994 •
1 answer
Personally I don't think that trying to align all 3'UTR of genes in a genome will result in anything decent! There is way too much variation (== too little conservation) to result in a sensible multiple alignment.
Moreover, simply doing a global alignment is probably technically not even feasible. Lengths of the 3'UTR are too variable and if you're looking for motifs those are too small (insignificant) to be visible in a global multiple alignment.
A better approach (= what most people will do ) is to first group genes (UTRs) in biological significant clusters (eg. co-expression, same pathway, ...) and then try to analyze the UTRs in those smaller groups. Even so, the simple MSA approach will then not even give much I assume. You're better trying software specifically for that purpose (= motif detection), I'm thinking of phylogenetic footprinting, phylogenetic shadowing, motif sampling, RSA-tools ....
Log in to answer this question.
You could try to use CD-Hit to remove redundancy from your sequences in order to reduce the data set size.