I have a genome, and each position has a conservation score. I want to identify the conserved fragments from these scores. A naive way would be to setup two thresholds, one for the conservation score at each base and another one could be the length of the fragments. I want to calculate a P-value or sort of things to quantify the conservation of a fragment. It sounds a very common task, for example, like the peak calling. So I am wondering whether anyone can recommend some methods to calculate such a score.
I highly appreciate your help!
•
written
by
alteralex •
For those that are interested in my question, I think I have found an answer from David Haussler's paper: http://genome.cshlp.org/content/15/8/1034.abstract
You might try Adam Siepel's phyloP and/or PhastCons. These provide (at least) a per-base conservation p-value, not yet what you want. But from glancing at the documentation (http://compgen.bscb.cornell.edu/phast/phastCons-HOWTO.html) it seems that the --most-conserved flag may detect highly conserved fragments. I've never used this feature myself, but it may be along the lines you're looking for. The per-base p-values are available on several UCSC conservation tracks.
Thank you bstephenwhite for your input. Indeed the score I have for each base is the PhastCons score. I want to base on these scores to find conserved fragments. And for genome-wide selection, I need a formal way to quantify the significance of the conservation of the fragments I found.
Log in to answer this question.
