FAST icon indicating copy to clipboard operation
FAST copied to clipboard

How can I use Deepfusion to predict the binding affinity of a new complex

Open NECCCCCCC opened this issue 4 years ago • 3 comments

Hi,

It seems that the code here can only evaluate the model on PDBBind. How can I use the code to predict the binding affinity of a new complex not in PDBBind? If you can provide a script or any other methods, it would be very helpful. Thanks.

NECCCCCCC avatar Jan 14 '22 03:01 NECCCCCCC

What format do you want to use for prediction? pdb/pdbqt/mol2?

We plan to release a data reader to support regular protein-ligand file formats such as mol2 and pdb. It would be slower but would be less hassle for conversion.

hkimdavis avatar Feb 01 '22 00:02 hkimdavis

It is great if Deepfusion can support any valid molecule file with pdb or mol2 format.

NECCCCCCC avatar Feb 14 '22 01:02 NECCCCCCC

Hi, Do you plan on publishing a binding affinity prediction tool that could be used as Vina? Based on your results it could be a much more reliable prediction tool.

tomcastigl avatar Jan 21 '23 17:01 tomcastigl