Inspiration
What it does
How we built it## Inspiration
As a computational biology student at UC Berkeley, I witnessed a critical pain point: many biomedical researchers lack coding skills to process multi-omics NGS data. Complex bioinformatics pipelines, confusing genomic annotation tools, and unreadable raw sequencing outputs slow down disease research and precision medicine. Inspired by OpenAI’s powerful language & code models, I built BioOmicsGPT to lower the technical barrier for omics analysis.
What it does
BioOmicsGPT is an AI assistant powered by OpenAI API that:
- Automatically annotates multi-omics sequencing datasets
- Generates ready-to-run Nextflow/Snakemake analysis pipelines
- Translates raw genomic results into plain-language biomedical reports
- Wraps complex bioinformatics scripts for non-technical lab researchers
How we built it
- Core logic built with Python, integrating OpenAI GPT-4 & Code Interpreter API
- Bioinformatics backend uses Bioconductor, Scanpy, SAMtools for omics processing
- Containerized via Docker for consistent HPC/cloud deployment
- Simple web frontend for uploading sequencing files and viewing AI-generated reports
Challenges we ran into
- Balancing accurate biological domain knowledge with general LLM outputs (mitigated by injecting curated omics prompt engineering)
- Optimizing API call cost & speed for large NGS dataset analysis
- Standardizing diverse multi-omics data formats for unified AI interpretation
What we learned
- Advanced prompt engineering for specialized life science domain tasks
- End-to-end integration of large language models with traditional bioinformatics workflows
- Designing AI tools that prioritize usability for non-computational domain experts
Log in or sign up for Devpost to join the conversation.