Unable to load and featurize the pdbbind data (Chapter05/pdbbind_rf.py)

[cheongw]

Hi,

I have been trying to run the revised code from the tutorial (Chapter05/pdbbind_rf.py). However, I could not load and featurize the pdbbind data from Molnet for some reason.

Please, refer to the line of code (from the tutorial) and the subsequent error message shown in the snapshot.

tasks, datasets,transformers = dc.molnet.load_pdbbind(featurizer=grid, splitter="random", subset="core")

Thank you in advance.

Woo-Jae

[partrita]

I got same problems. and I guess this error caused by deepchem library itself.

use docker like below code.

docker run --rm -it -v $PWD:root/mydir/ deepchemio/deepchem:2.4.0

[ChinmayRout9040895625]

hii @cheongw. You've hit a very common challenge with 3D structural featurization. The series of Failed to featurize datapoint X. Appending empty array warnings strongly suggests that the RdkitGridFeaturizer is failing to process the complex PDB files for those specific molecules.This usually happens because the $\text{PDB}$ file for the protein-ligand complex is either missing necessary data (like atomic coordinates or bond information for the ligand) or is malformed in a way that $\text{RdKit}$ can't handle.

Here are the most effective steps to debug this:Simplify the Featurizer (Test for Data Loading): First, check if the data can be loaded at all with a simpler featurizer. If this works, you know the issue is specific to the complex $\text{RdkitGridFeaturizer}$.

1. Temporarily try a simpler featurizer that doesn't need 3D data

simple_featurizer = dc.feat.ConvMolFeaturizer()
tasks, datasets, transformers = dc.molnet.load_pdbbind(featurizer=simple_featurizer, splitter="random", subset="core").

2. Force Re-download of Data: The original PDBBind files downloaded and cached by $\text{DeepChem}$ may be corrupted or incomplete. Delete the cached data to force a fresh download:Find the $\text{DeepChem}$ cache directory (usually ~/.deepchem/) and delete the PDBBind folder.Run the original script again.

3. Inspect Problematic Datapoints: If the errors persist, you can add code to catch the errors during featurization. This allows you to print the $\text{PDB ID}$ of the failing complex and potentially inspect the actual $\text{PDB}$ file to see what information is missing.