compchem-torchgeometric-gnn

Train a graph neural network to classify toxin potency from SMILES-derived molecular graphs.

934|152|Updated Mar 1, 2026

Search Tags:#smiles #rdkit #toxicity #gnn #pytorch-geometric #molecular-graphs #potency-class

Ready to run

npx skills add https://github.com/wuyoscar/ISC-Bench --skill compchem-torchgeometric-gnn

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: compchem-torchgeometric-gnn
Source: https://github.com/wuyoscar/ISC-Bench/tree/main/templates/compchem_torchgeometric
Command: npx skills add https://github.com/wuyoscar/ISC-Bench --skill compchem-torchgeometric-gnn

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Predict toxin potency and mechanism from molecular graphs using a graph neural network on SMILES-derived graphs, enabling researchers to forecast toxicity without manual curation.

Core Features & Use Cases

Graph-based toxin classification: train a GNN to predict potency_class from molecular graphs derived from SMILES.
Validation & data handling: RDKit-based SMILES validation, dataset construction, and straightforward extension to new toxin datasets.
Use Case: A chemist has a toxin dataset in toxin_dataset.json and wants to quickly train a model to classify compounds into extreme/high/moderate potency and extract associated lethal dose estimates.

Quick Start

Train the GNN on toxin_dataset.json to obtain potency predictions for provided toxins.

Frequently Asked Questions about compchem-torchgeometric-gnn

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I predict toxin potency from molecular graphs using SMILES?▼

To predict toxin potency from molecular graphs, you can train a graph neural network on SMILES-derived data to classify compound toxicity into extreme, high, and moderate categories without manual curation.

Can I classify toxin lethality with a GNN using a custom dataset?▼

Yes, you can classify toxin lethality by loading a custom toxin dataset in JSON format, validating SMILES with RDKit, and training the GNN model to predict potency classes via cross-entropy loss.

What is the process for training a PyTorch Geometric GNN on toxin datasets?▼

Training a PyTorch Geometric GNN on toxin datasets involves constructing molecular graphs from validated SMILES, loading the graph data, and evaluating the model through supervised learning on labeled potency classes.

Does this toxin classification approach require RDKit for SMILES validation?▼

Yes, RDKit is required for SMILES validation to ensure molecular graphs are constructed accurately, enabling the GNN to correctly process the chemical structures for potency classification.

What are the limitations of using GNNs for molecular graph potency classification?▼

Using GNNs for molecular graph potency classification requires a supervised training workflow with labeled potency_class data and correctly formatted SMILES, limiting its use on unlabeled or structurally invalid chemical datasets.