Skip to content

GeneGround

GeneGround is a claim-level grounding layer for AI-generated omics interpretations.

Code Video Cerebral Valley

Core trick

Splits an omics summary into atomic claims, retrieves matching evidence chunks, and lets specialist agents judge each claim before deterministic aggregation sets the verdict.

Tags

scienceagents

Description

GeneGround is a claim-level grounding layer for AI-generated omics interpretations. I built it to address a core problem in AI-assisted biology: LLMs can generate fluent scientific narratives, but they may subtly overstate causality, mechanism, robustness, or biological scope. GeneGround takes an AI-generated omics interpretation and evidence handoff file, then decomposes the interpretation into individual biological claims. It normalizes genes, pathways, cell contexts, conditions, and directions against controlled taxonomies; indexes the handoff into perturbation, pathway, robustness, provenance, and language-rule evidence; retrieves relevant chunks using metadata-first retrieval with local TF-IDF fallback; and evaluates each claim with four specialist Claude-powered agents. Final verdicts are assigned through deterministic aggregation, while Claude is used for semantic biological judgment and grounded rewrites. The Technical Pipeline page exposes every processing stage from claim extraction through retrieval and verdict generation, while an interactive Evidence Trace links each sentence to its claims, agent queries, retrieved evidence chunks, and final reasoning. In the demo, GeneGround audited a primary human CD4+ T cell Perturb-seq interpretation, identifying claims supported with caveats, partially supported, or overstated. This matters because AI-assisted science needs traceable evidence, cautious language, and reproducible reasoning, not just plausible summaries.
rheazhou2026/geneground★ 10 forksTypeScriptMITCreated Jul 11, 2026 · last push Jul 14, 2026

README excerpt

# GeneGround

**A claim-level grounding checker for AI-assisted genomics interpretation.**

GeneGround decomposes AI-generated omics summaries into individual biological claims, checks each one against the underlying dataset and published literature, and returns an auditable, evidence-linked verdict — so a scientist can trust an AI interpretation before it shapes a paper, presentation, or downstream experiment.

Built for the **Built with Claude: Life Sciences** hackathon (Builder track), powered by **Claude Code** and **Claude Science**, and demoed on the Gladstone/UCSF T cell Perturb-seq dataset.

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](./LICENSE)
[![Built with Claude](https://img.shields.io/badge/Built%20with-Claude-6A5ACD)](https://www.anthropic.com/news/claude-science-ai-workbench)
[![Next.js](https://img.shields.io/badge/Next.js-000000?logo=next.js&logoColor=white)](https://nextjs.org/)
[![TypeScript](https://img.shields.io/badge/TypeScript-3178C6?logo=typescript&logoColor=white)](https://www.typescriptlang.org/)
[![Deployed on Vercel](https://img.shields.io/badge/Deployed%20on-Vercel-black?logo=vercel)](https://geneground.vercel.app)

**[Live Demo](https://geneground.vercel.app)** · **[Demo Video](https://www.youtube.com/watch?v=aILeQ9WDXzs)** · **[Report an Issue](https://github.com/rheazhou2026/geneground/issues)**

---

## Table of Contents

- [The Problem](#the-problem)
- [What GeneGround Does](#what-geneground-does)
- [Key Features](

Team