NaNs in PSSMs
Closed this issue · 2 comments
aarmey commented
Why do you end up with a ton of NaN's in your pssms? I'm seeing tons of warnings when I build the figures.
mcreixell commented
I just checked and I'm not seeing any NaNs in the PSSMs (pssms = model.pssms(PsP_background=False/True)
). I think that the warnings may be due to the central position in which all are 0 except from Y, S, T?
aarmey commented
Aha. Those probably turn into -inf after log transforming. We can discuss this next time we're on Zoom. There's likely a quick fix.