Drug Molecular Toxicity Prediction using Deep Neural Networks
- 0 Collaborators
All human beings are exposed to different chemicals during their lifetime through food, medicines, daily-life products etc. over 30% of drugs have failed in human clinical trials because they are determined to be toxic despite promising pre-clinical studies in animal models. Real-world chemical trials for assessing drugs are extremely time consuming. It will be ideal if a computational drug molecular toxicity assessment method can be developed to test the chemical's toxicity. ...learn more
Project status: Under Development
Overview / Usage
The project will not only help in effective prediction of molecular toxicity but also save huge amounts of time and resources that is spent in molecule toxicity assessment.
Methodology / Approach
As deep learning has shown advantage in handling large amount of data and is able to make predictions based on the data, we will be using it to predict the toxicity of chemical molecules using their SMILES structure. Simplified Molecular-Input Line-Entry System (SMILES) is a linear representation for molecular structure using 1D ASCII strings. We first convert the SMILES structure into one-hot representation vector to feed it into our Neural Network. Currently, we are using a very simple CNN model to predict the toxicity. Simultaneously, we are trying to find and develop a better neural network for more accurate predictions and better performance.
Technologies Used
Numpy
Python
Tensorflow
Repository
https://github.com/muditchaudhary/DrugMolecularToxicityPrediction