graphein icon indicating copy to clipboard operation
graphein copied to clipboard

Structure-informed dataset splitting

Open a-r-j opened this issue 5 years ago • 0 comments
trafficstars

Create good train/val/test sets based on SCOP/CATH classifications. Sequence-based approaches (e.g. identity thresholding or BLAST) are bad practice and should not be encouraged.

a-r-j avatar Jul 16 '20 09:07 a-r-j