Entering edit mode
9.8 years ago
by0
▴
110
Hi,
I am constructing my own position weight matrices (PWMs) and scoring regions of the yeast genome, for a list of predicted TFBS. I am then using different features to filter this list, in effort to minimize the false positive rate.
My question is, since there is no experimentally defined functional tfbs in yeast (at least that I know of), what is the best way to benchmark my features?
Thank you!
What do you mean there are no experimentally verified tfbs in yeast? There are whole databases of them if you google a bit.
Let me clarify, I meant functional tfbs. My filtering is in effort to identify functional tfbs. I will update the question
If there really aren't any (I find that difficult to believe, but I suppose it could happen), then why not just validate the sites for at least one transcription factor? There's at least one published method for this using a one-hybrid approach. Then you'd have the known positives you need.
None that are annotated and readily accessible from a database in the public domain for yeast. Thanks for the tip though.