# Understanding protein function with a multimodal retrieval-augmented foundation model

arXiv: https://arxiv.org/abs/2508.04724 · HTML: https://mlsanity.com/paper/2508.04724/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in Medical and biomedical RAG, one of 15 directions in Retrieval-augmented generation.
- Builds on 5 of its 191 references that are in the map.
- First posted on arXiv 5 Aug 2025.

## Placed in

- Retrieval-augmented generation / [Medical and biomedical RAG](https://mlsanity.com/rag/medical-and-biomedical-rag/)

## Builds on (most cited first, 5 of 5)

- [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://mlsanity.com/paper/2104.09864/) (2021)
- [GLU Variants Improve Transformer](https://mlsanity.com/paper/2002.05202/) (2020)
- [Learning from Protein Structure with Geometric Vector Perceptrons](https://mlsanity.com/paper/2009.01411/) (2020)
- [Multi-Scale Representation Learning for Protein Fitness Prediction](https://mlsanity.com/paper/2412.01108/) (2024)
- [Retrieval-Enhanced Mutation Mastery: Augmenting Zero-Shot Prediction of Protein Language Model](https://mlsanity.com/paper/2410.21127/) (2024)

## Built on by (most cited first, 6 of 6)

- [EvoFlows: Evolutionary Edit-Based Flow-Matching for Protein Engineering](https://mlsanity.com/paper/2603.11703/) (2026)
- [SciDocBench: A Workflow-Centered Benchmark and Data Pipeline for Scientific Document Understanding](https://mlsanity.com/paper/2609.05141/) (2026)
- [No Generation without Representation: Efficient Causal Protein Language Models Enable Zero-Shot Fitness Estimation](https://mlsanity.com/paper/2602.01845/) (2026)
- [Pareto-Optimal Offline Reinforcement Learning via Smooth Tchebycheff Scalarization](https://mlsanity.com/paper/2604.13175/) (2026)
- [MIMIC: A Generative Multimodal Foundation Model for Biomolecules](https://mlsanity.com/paper/2604.24506/) (2026)
- [Flexible Flows for Biological Sequence Design](https://mlsanity.com/paper/2606.10543/) (2026)

## Abstract (arXiv)

Protein language models (PLMs) learn probability distributions over natural protein sequences. By learning from hundreds of millions of natural protein sequences, protein understanding and design capabilities emerge. Recent works have shown that scaling these models improves structure prediction, but does not seem to improve mutation understanding and representation quality for protein function prediction. We introduce PoET-2, a multimodal, retrieval-augmented protein foundation model that incorporates in-context learning of family-specific evolutionary constraints with optional structure conditioning to learn generative distributions over protein sequences. PoET-2 uses a hierarchical transformer encoder that is equivariant to sequence context ordering and a dual decoder architecture with both causal and masked language modeling objectives, allowing PoET-2 to operate in both fully generative and bidirectional representation learning modes. PoET-2 achieves state-of-the-art performance on zero-shot variant effect prediction, excelling at scoring variants with multiple mutations and challenging indel mutations. In supervised settings, PoET-2 embeddings outperform previous methods for learning sequence-function relationships, especially with small datasets. This work highlights the benefits of combining retrieval augmentation with multimodal, family-centric modeling for advancing protein foundation models.