Prediction of Drug - Target Interaction with Interpretable Nested Graph Neural Network and Pretrained Molecule Models (iNGNN-DTI)

Introduction

iNGNN-DTI is a framework for the drug-target interaction prediction. The model extracts features of from the graph data of drugs and targets, employing a specific type of graph neural network known as the nested graph neural network (NGNN), in which the target graph is created using Alphafold2. We use the attention-free transformer (AFT) module to capture the interaction information between the substructures of drugs and targets. To improve the feature representations, we integrate features learned by models that are pre-trained on large unlabeled small molecule and protein datasets for the drugs and targets, respectively.

Environment

We conduct our experiments with python3.8. Here are the requirements

descriptastorus
matplotlib
networkx
numpy
pandas
prettytable
rdkit
Requests
scikit_learn
scipy
subword_nmt
torch
torch_geometric
torchvision

Usage

python main.py

Acknowledgement

DeepPurpose: https://github.com/kexinhuang12345/DeepPurpose
Nested Graph Neural Network (NGNN): https://github.com/muhanzhang/NestedGNN
Attention Free Transformer: https://github.com/rish-16/aft-pytorch