Welcome to PSICHIC Server!
PSICHIC (PhySIcoCHemICal graph neural network) is a groundbreaking tool for drug discovery, predicting protein-ligand interactions directly from sequence data. Here's why it's so special:
- State-of-the-Art: Trained on identical protein-ligand pairs without structural data, PSICHIC matched and even surpassed state-of-the-art structure-based methods in binding affinity prediction. PSICHIC also achieved a high 0.96 accuracy in predicting functional effects or in other words, the way in which the drug might affect our bodies.
- Emergent Interpretability: PSICHIC's interpretable fingerprint decodes mechanisms underlying protein-ligand interactions, identifying binding site protein residues and involved ligand atoms from sequence data alone.
- Experimentally Validated: Model capabilities have also been experimentally validated to discover potential drugs.
How to Use PSICHIC (sequence data is all you need):
- Step 1: Enter a protein sequence.
- Step 2: Enter ligand's SMILES string(s).
- Step 3: Run the model and discover your next big breakthrough!
NOTE: We only allow ONE protein target against up to 1,000 ligand compounds per run. Estimated completion time is ~1 minute. We are hosting potentially many jobs with only one server, and each run needs to reinitialize the model from scratch, so it could take longer.
For more flexible and scalable processing (up to 100K compounds in an hour), use the free GPU on Colab
or deploy locally using
GitHub
.
For more flexible and scalable processing (up to 100K compounds in an hour), use the free GPU on Colab
or deploy locally using
GitHub
.