Can Biomni analyze my own research data?

Yes, you can provide file paths for local datasets, such as RNA-seq h5ad files or GWAS summary statistics, for the agent to autonomously process and analyze.

Which LLM models are recommended for use with Biomni?

Claude 3.5 Sonnet is the recommended model for its balance of speed and biological reasoning quality, though it also supports GPT-4, Gemini, and Llama 3.

What research domains does Biomni support?

Biomni supports a wide range of fields including genomics, proteomics, drug discovery, clinical analysis, molecular biology, and laboratory protocol optimization.

How does Biomni handle code execution security?

Biomni executes LLM-generated code with system privileges. It is strongly recommended to run it in isolated environments like Docker or VMs when processing sensitive data.

Does Biomni require external databases?

Biomni includes an integrated ~11GB data lake containing gene databases (Ensembl, NCBI), protein structures (PDB, AlphaFold), clinical datasets (ClinVar, OMIM), and literature indices.

Biomni Biomedical Agent

Name: Biomni Biomedical Agent
Author: Microck

byMicrock

•

Data Science & ML

Executes complex autonomous biomedical research tasks including genomics, drug discovery, and clinical data analysis.

About

Biomni is an open-source framework from Stanford's SNAP lab designed to function as an autonomous AI agent for biomedical research. It bridges the gap between large language model reasoning and specialized biological data by integrating roughly 11GB of biomedical databases, literature, and ontologies. The skill enables users to automate sophisticated workflows such as CRISPR screening design, single-cell RNA-seq analysis, and ADMET prediction for drug discovery. By combining dynamic code execution with deep domain knowledge, Biomni streamlines multi-step biological reasoning and complex data interpretation within a unified, programmable interface.

Key Features

Integrated access to 11GB of curated biomedical databases and literature
Dynamic code generation for genomics and proteomics analysis pipelines
Cross-domain support for drug discovery, GWAS, and clinical genomics
Built-in evaluation framework for benchmarking biomedical task performance
81 GitHub stars
Autonomous multi-step biological reasoning and task decomposition

Use Cases

Predicting ADMET properties and identifying potential targets for drug discovery
Designing and prioritizing genes for genome-wide CRISPR knockout screens
Automating single-cell RNA-seq pipelines including clustering and cell annotation

About

Key Features

Integrated access to 11GB of curated biomedical databases and literature
Dynamic code generation for genomics and proteomics analysis pipelines
Cross-domain support for drug discovery, GWAS, and clinical genomics
Built-in evaluation framework for benchmarking biomedical task performance
81 GitHub stars
Autonomous multi-step biological reasoning and task decomposition

Use Cases

Predicting ADMET properties and identifying potential targets for drug discovery
Designing and prioritizing genes for genome-wide CRISPR knockout screens
Automating single-cell RNA-seq pipelines including clustering and cell annotation