Predict molecular, protein, knowledge graph, and retrosynthesis outcomes from graph-structured bioinformatics data.
npx skills add https://github.com/dralkh/seerai --skill torchdrug-dralkh
Please help me install this Agent Skill. Skill: torchdrug Source: https://github.com/dralkh/seerai/tree/main/skills/torchdrug Command: npx skills add https://github.com/dralkh/seerai --skill torchdrug-dralkh
TorchDrug removes the complexity of building graph-based machine learning workflows for chemistry and biology, helping you predict properties, reason over knowledge graphs, generate molecules, and plan syntheses from a unified toolkit.
Use the torchdrug skill to set up a graph learning workflow for your dataset, select the recommended model and task, and evaluate it with the appropriate metrics and split strategy.
High-intent search queries and answers about installing and using this skill.
Predict molecular properties by building graph neural network classifiers and regressors for ADMET, toxicity, and solubility. You configure PyTorch graph models, apply scaffold splits, and evaluate outcomes using metrics like AUROC and AUPRC.
Yes, you can perform retrosynthesis by decomposing synthesis planning into reaction-center identification and synthon completion. Knowledge graph reasoning is supported for link prediction on biomedical graphs, evaluated using metrics such as MRR.
Protein modeling predicts enzyme function, localization, stability, and protein interactions from sequence or structure data. You select a strong baseline model, configure task-specific losses, and evaluate the predictions using appropriate metrics.
Yes, the workflow requires configurable PyTorch graph models and curated datasets like BBBP, EnzymeCommission, or Hetionet. You select the recommended model and task-specific losses to quickly set up a publishable experiment.
Binding prediction and molecular generation tasks are evaluated using metrics such as AUROC, AUPRC, and MRR. You select the appropriate metric set and dataset split strategy to measure the performance of your graph models.
A unified graph learning toolkit removes the complexity of building machine learning workflows for chemistry and biology. It allows you to predict properties, reason over knowledge graphs, generate molecules, and plan syntheses from a single framework.
biomaps-infraDevelop drug discovery AI models with PyTorch-native graph neural networks.
Zeyuyang-0420Predict molecular properties and model proteins using graph neural networks.
shushuznTrain PyTorch graph neural networks for molecular and protein property prediction.
galeepPredict molecular properties and model proteins with custom GNN architectures.
