Research map / Vision-language and multimodal / Cross-modal learning
Darwin-7B: A Multi-Omic Foundation Model for the Human Gut Microbiome via Sparsified Quality-Aware Tokenization
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes Darwin-7B, a multi-omic foundation model for the human gut microbiome that jointly models metagenomic and metabolomic content via sparsified, quality-aware tokenization.
Problem
Existing biological foundation models operate on a single modality and on assembled single-organism genomes, and so cannot represent the joint metagenomic and metabolomic state of microbial ecosystems.
Approach
A foundation model over paired metagenomic and metabolomic profiles of microbiomes, using sparsified quality-aware tokenization to represent community-level multi-omic state rather than single-organism genomes.
Closest papers by meaning
- Scaling Laws and Architectural Frontiers in Metagenomic Foundation Models (ICML 2026 · similarity 0.89)
- Whole Genome Transformer for Gene Interaction Effects in Microbiome Habitat Specificity (AAAI 2025 · similarity 0.87)
- DNABERT-2: Efficient Foundation Model and Benchmark For Multi-Species Genomes (ICLR 2024 · similarity 0.84)
- FunctionEvolve: Structure-Guided Symbolic Regression with LLMs (NeurIPS 2026 · similarity 0.83)
- Training Compute-Optimal Protein Language Models (NeurIPS 2024 · similarity 0.83)
- What DNA Foundation Models Learn Beyond Sequence Composition (NeurIPS 2026 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.