Tatta Bio provides an AI‑driven platform that unifies large‑scale genomic and protein sequence data with advanced language models to predict protein functions and interactions. Their SeqHub web interface and tools like FlashPPI and SeqHub Agent enable rapid, linear‑time annotation and interaction mapping, while open datasets and model code lower barriers for high‑throughput genomic analysis.
Funding
Funding not disclosed
Founders
Product
Problem
Researchers struggle to interpret the vast amount of genomic and protein sequence data because existing tools are fragmented, computationally intensive, and provide limited functional annotation for the majority of sequences.
Solution
Tatta Bio develops AI-driven genomic intelligence platforms that unify large-scale sequence datasets, advanced language models, and interactive tools to predict protein functions and interactions. Their mixed-modality genomic language model (gLM2) learns contextual representations from both coding and intergenic regions, enabling accurate functional embeddings. The SeqHub web platform integrates these embeddings with functional annotations, allowing rapid, linear‑time protein‑protein interaction prediction (FlashPPI) and AI‑assisted gene annotation (SeqHub Agent). By providing open datasets such as the OMG corpus and publicly available model code, Tatta Bio lowers the barrier for researchers to perform high‑throughput, AI‑enhanced analysis of microbial and other genomes.
Target Audience
Primary users are academic and industry researchers in genomics, microbiology, and protein science who need scalable AI tools for functional annotation and interaction mapping of biological sequences.
Features
- gLM2 mixed‑modality language model that captures genomic context and residue‑level co‑evolution signals
- FlashPPI contrastive learning framework for linear‑time proteome‑wide protein interaction prediction
- SeqHub interactive web platform combining predicted networks, functional annotations, and genomic context
- SeqHub Agent (formerly Gaia Agent) AI biologist that annotates uncharacterized genes using bioinformatics tools and contextual reasoning
- Open OMG corpus (3.1 Tbp, 3.3 B coding sequences) for pre‑training and benchmarking genomic models
- Publicly released model code and datasets via GitHub for reproducibility and community use