Deepfrag output seems to teleport ligand to (0,0,0)

[Ben Cree]

Hello,

Certain ligands when grown with Deepfrag seem to be moved to near the origin (coordinates (0,0,0), at least the top few output molecules by score do, I have not checked them all). Initially I thought it was something to do with code I wrote – and so tested my input files with the web app, and the same issue seems to occur. Is there a way to avoid this?

The input files I was using are available here : https://github.com/BenCree/deepfrag
The growing point I selected was the carbon ortho to the carboxylate group, and an example of the translated output is ‘fused(5).sdf’.

Cheers,
Ben

[Jacob Durrant]

Hi Ben. Much thanks for your interest in DeepFrag! Just to clarify, DeepFrag doesn’t actually position the fragments it recommends in 3D space. It only suggests which fragments to add.

As a helper program for the DeepFrag browser app, we created a separate app called Fuser that can be used to add the DeepFrag-recommended fragment to a source molecule. The output is a 3D model for use in docking studies. Are you referring to the Fuser app in your question, by chance?

If so, it’s important to note that the 3D ligand made in Fuser isn’t meant to be in any sense "docked" into the protein receptor. DeepFrag doesn’t position the fragment (it only recommends what fragment to use), and Fuser is a step removed from DeepFrag.

Hope this helps. Take care..

[Author]

Posts