Closed SeppeDeWinter closed 4 weeks ago
Note, code for motif embedding has not been tested yet.
Okay, looks good and makes sense to me. In the near future we should really make a separate tutorial for enhancer design (including this information here), since as of now it's a one-liner in the introductory tutorial. @erceksi could you take a look too since you implemented the original function?
Added some extra changes.
Now multiple sequences should be processed in parallel.
Before a call was made to model.predict
for each sequence and each iteration.
Now a single call is made to model.predict
for each iteration only.
From a quick and dirty benchmark, this code should be around 2x faster.
For example: