harragi/CellAg

CellAg is an open research workspace that applies AI/ML and foundation models to cellular agriculture problems (media cost, scale, scaffold design, sensory prediction). It’s aimed at researchers and engineers designing and validating cell-culture media and sensory outcomes. Key features: - A runnable interactive agent (Claude Agent SDK) that integrates KEGG, Ensembl, EuropePMC, arXiv and a curated cell‑ag bibliography to drive media design (Project Sardine). - Unified composition (Sardine) and validation (Hamster) workflows: propose media formulations, run iCHO-model validation and metabolic yield calculations, and produce editable notes. - Web UI (React) + Bun/TypeScript server; custom tools (read/propose notes edits, query_kegg/ensembl/europepmc/arxiv/bigg/chebi, compute yields). - Interactive benches: cho-bench (Monod-style ODE triage for CHO media) and sensory-bench (flavor prediction via Maillard, lipid oxidation, umami models). - Propose-then-apply editing model; nothing is written without user confirmation. Requires Bun and an Anthropic API key.

このRepoのdescription / GitHub topics / READMEのAI要約に出現したAI関連キーワードをカテゴリ別に表示。各バッジは該当カテゴリの詳細ランキングへリンクします。