chemberta

This project presents a Graph Neural Network (GNN)-based framework for predicting Drug-Drug Interactions (DDIs) using pretrained SMILES embeddings and Graph Attention Networks (GAT). Given the limitations of clinical studies and traditional computational methods in detecting complex DDIs.

Physics-informed machine learning and cheminformatics workflow for predicting biochar adsorption capacity of organic water contaminants using RDKit descriptors, molecular fingerprints, ChemBERTa embeddings, and a Streamlit research app.

Deep Learning project for Drug-Drug Interaction (DDI) Prediction using ChemBERTa, molecular SMILES representations, and Hugging Face Transformers.

AI-powered multi-task framework for predicting cardiac ion channel toxicity using GNNs, ChemBERTa, and machine learning to support safer drug discovery.

Live SMILES-to-toxicity screening dashboard comparing 5 ML model families (stacked ensembles, Tanimoto GPC, autoencoder, ChemBERTa) with consensus scoring — best external ROC-AUC 0.860.

This is a public repository for the Kaggle competition, Thermophysical Property: Melting Point. This project leverages ChemBERT to obtain SMILES embeddings for downstream regression models.

Molecular ML pipeline for ADMET prediction using RDKit, Morgan fingerprints, ChemBERTa embeddings, and SLURM-based HPC workflows.

ML pipeline predicting small-molecule binding affinity and mutant-vs-wild-type selectivity for the KIT kinase in mast cell disease.