Open WeiCSong opened 10 months ago
Sorry for the very late follow-up, no need to reverse DNA sequences as the model by default passes in both the input sequence and the reverse complement, and averages these predictions (that is, your intuition is correct that Sei does not have strand-specificity)
Hi, i'm wondering if it is neccessary to reverse all input DNA sequence to positive strand when using sei. Biologically speaking both strand should have the same peak but i'm not sure whether this is guaranteed in the training. Thanks for your help!