
from awesome-bio-agent-skills66
Authoritative guide for building and scaling Graph Neural Networks using PyTorch Geometric: data, layers, training patterns, and common pitfalls.
This skill equips an agent with practical knowledge to build, train, and scale Graph Neural Networks using PyTorch Geometric (PyG). It explains core data structures (Data, HeteroData), dataset usage, transforms, common convolution layers, message-passing patterns, and strategies for neighbor sampling and multi-graph batching.
Trigger when a user asks about graph learning, node/graph classification, link prediction, heterogeneous graphs, or when code imports torch_geometric. Use during model development, debugging GNN architectures, or adapting GNNs to large graphs that require sampling or cluster-based loaders.
Well-suited for code-capable assistants and notebooks (Copilot/Codex, Claude Code, Jupyter/Colab helpers) that can provide code snippets, debugging help, and model training guidance.
This skill has not been reviewed by our automated audit pipeline yet.
arXiv Search
Search arXiv for recent preprints and build local Markdown summaries; ideal for CS, math, physics, and quantitative-bio literature discovery.
Alternative Splicing Analysis Pipeline
End-to-end pipeline for short-read bulk RNA-seq alternative splicing analysis: QC, STAR 2-pass alignment, junction QC, rMATS differential splicing, isoform swit
PLINK Basics
Commands and examples for converting genotype formats and running standard QC filters (MAF, geno, mind, HWE) with PLINK 1.9/2.0.
AlphaFold Structure Query
Query the AlphaFold EBI API to fetch predicted protein structures, download PDB/CIF files, and inspect confidence metrics (pLDDT/PAE).