Interpreting Protein Language Model Embeddings via Orthogonal Projection for Protein Fitness Prediction

• URL: http://arxiv.org/abs/2608.25548v1
• Date: Wed, 26 Aug 2026 09:01:43 GMT
• Title: Interpreting Protein Language Model Embeddings via Orthogonal Projection for Protein Fitness Prediction
• Abstract summary: We show that PLM embeddings encode patterns correlated with biochemical properties and quantify their contribution to predicting protein fitness.<n>This computationally efficient approach is not limited to the features or embeddings considered here and is readily transferable to problem settings beyond protein fitness prediction.
• Score: 2.365650942022388
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Recently, there has been a growing adoption of protein language models (PLMs) in biomedical science. Their embeddings provide a rich numerical representation of protein sequences which achieve state-of-the-art performance on several downstream tasks including protein fitness prediction. However, PLM embeddings are not directly interpretable and, thereby, it remains unclear what features they encode. To gain insight into which biochemical properties of the protein are driving the prediction, we leverage an orthogonal projection technique that removes linear effects of known tabular features from embeddings and extend it to high-order and interaction effects. In this way, we remove the effects of interpretable biochemical features from PLM embeddings. In an ablation study, we show that this leads to a decrease in performance for a downstream classifier trained only on the embeddings to predict protein fitness. In an additional evaluation, we find that these biochemical features explain a substantial part of the variance in the predictions of this classifier. Hence, we can show that PLM embeddings encode patterns correlated with biochemical properties and quantify their contribution to predicting protein fitness. This computationally efficient approach is not limited to the features or embeddings considered here and is readily transferable to problem settings beyond protein fitness prediction.

Related papers

• Sparse Autoencoders for Low-$N$ Protein Function Prediction and Design [0.0]
  Predicting protein function from amino acid sequence remains a central challenge in data-scarce regimes.<n>Protein language models (pLMs) have advanced the field by providing evolutionary-informed embeddings and sparse autoencoders (SAEs)<n>We show that SAEs, with as few as 24 sequences, consistently outperform or compete with their ESM2 baselines in fitness prediction.
  arXiv  Detail & Related papers  (2025-08-25T23:56:39Z)
• Protein Large Language Models: A Comprehensive Survey [71.65899614084853]
  Protein-specific large language models (Protein LLMs) are revolutionizing protein science by enabling more efficient protein structure prediction, function annotation, and design.<n>This work provides the first comprehensive overview of Protein LLMs, covering their architectures, training datasets, evaluation metrics, and diverse applications.
  arXiv  Detail & Related papers  (2025-02-21T19:22:10Z)
• Computational Protein Science in the Era of Large Language Models (LLMs) [54.35488233989787]
  Computational protein science is dedicated to revealing knowledge and developing applications within the protein sequence-structure-function paradigm.<n>Recently, Language Models (pLMs) have emerged as a milestone in AI due to their unprecedented language processing & generalization capability.
  arXiv  Detail & Related papers  (2025-01-17T16:21:18Z)
• Long-context Protein Language Modeling Using Bidirectional Mamba with Shared Projection Layers [76.95505296417866]
  Self-supervised training of language models (LMs) has seen great success for protein sequences in learning meaningful representations and for generative drug design.<n>Most protein LMs are based on the Transformer architecture trained on individual proteins with short context lengths.<n>In this work, we propose LC-PLM based on an alternative protein LM architecture, BiMamba-S, built upon selective structured state-space models.
  arXiv  Detail & Related papers  (2024-10-29T16:43:28Z)
• Protein-Mamba: Biological Mamba Models for Protein Function Prediction [18.642511763423048]
  Protein-Mamba is a novel two-stage model that leverages both self-supervised learning and fine-tuning to improve protein function prediction. Our experiments demonstrate that Protein-Mamba achieves competitive performance, compared with a couple of state-of-the-art methods.
  arXiv  Detail & Related papers  (2024-09-22T22:51:56Z)
• ProtT3: Protein-to-Text Generation for Text-based Protein Understanding [88.43323947543996]
  Language Models (LMs) excel in understanding textual descriptions of proteins. Protein Language Models (PLMs) can understand and convert protein data into high-quality representations, but struggle to process texts. We introduce ProtT3, a framework for Protein-to-Text Generation for Text-based Protein Understanding.
  arXiv  Detail & Related papers  (2024-05-21T08:06:13Z)
• ProLLM: Protein Chain-of-Thoughts Enhanced LLM for Protein-Protein Interaction Prediction [54.132290875513405]
  The prediction of protein-protein interactions (PPIs) is crucial for understanding biological functions and diseases. Previous machine learning approaches to PPI prediction mainly focus on direct physical interactions. We propose a novel framework ProLLM that employs an LLM tailored for PPI for the first time.
  arXiv  Detail & Related papers  (2024-03-30T05:32:42Z)
• NaNa and MiGu: Semantic Data Augmentation Techniques to Enhance Protein Classification in Graph Neural Networks [60.48306899271866]
  We propose novel semantic data augmentation methods to incorporate backbone chemical and side-chain biophysical information into protein classification tasks. Specifically, we leverage molecular biophysical, secondary structure, chemical bonds, andionic features of proteins to facilitate classification tasks.
  arXiv  Detail & Related papers  (2024-03-21T13:27:57Z)
• Multi-level Protein Representation Learning for Blind Mutational Effect Prediction [5.207307163958806]
  This paper introduces a novel pre-training framework that cascades sequential and geometric analyzers for protein structures. It guides mutational directions toward desired traits by simulating natural selection on wild-type proteins. We assess the proposed approach using a public database and two new databases for a variety of variant effect prediction tasks.
  arXiv  Detail & Related papers  (2023-06-08T03:00:50Z)
• Protein Representation Learning by Geometric Structure Pretraining [27.723095456631906]
  Existing approaches usually pretrain protein language models on a large number of unlabeled amino acid sequences. We first present a simple yet effective encoder to learn protein geometry features. Experimental results on both function prediction and fold classification tasks show that our proposed pretraining methods outperform or are on par with the state-of-the-art sequence-based methods using much less data.
  arXiv  Detail & Related papers  (2022-03-11T17:52:13Z)
• Leveraging Sequence Embedding and Convolutional Neural Network for Protein Function Prediction [27.212743275697825]
  Main challenges of protein function prediction are the large label space and the lack of labeled training data. Our method leverages unsupervised sequence embedding and the success of deep convolutional neural network to overcome these challenges.
  arXiv  Detail & Related papers  (2021-12-01T08:31:01Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.