Antifreeze peptides inhibit ice crystal growth and recrystallization, and are promising components of cryoprotective formulations for cell, tissue, and food preservation, as well as anti-icing surface coatings. However, the discovery of new antifreeze peptides has been hindered by their sequence diversity and the limited scalability of experimental screening. In this study, we identify novel antifreeze peptide candidates from a microbiome-derived sequence library using ensemble machine learning and molecular dynamics (MD) simulations. We developed an ensemble classifier composed of 10 adapter-tuned protein-language models and a random forest meta-learner. After training on a curated dataset of 73 766 sequences, we applied this ensemble to 56 008 amino acid sequences from an Arctic microbiome library to identify antifreeze peptide candidates. Structural prediction yields a diverse range of conformations for six selected candidates, including α-helices, coils, and combinations of both. To evaluate their functional relevance, atomistic MD simulations were conducted to assess conformational stability and solvent interactions under freezing conditions. One candidate shows persistent helicity, surface amphipathicity, and an organized hydration pattern consistent with structural signatures reported for ice-binding helices. These findings expand the known landscape of antifreeze peptides and highlight a scalable strategy for discovering functional peptides from complex biological sources.

Subcellular localization prediction is crucial for understanding protein functions and cellular processes. Subcellular localization is dependent on tissue and cell lines derived from different cell types. Predicting cell line-specific subcellular localization using the information of protein-protein interactions (PPIs) offers deeper insights into dynamic cellular organization and molecular mechanisms. However, many existing PPI networks contain systematic errors that limit prediction accuracy. In this study, we propose a reinforcement learning approach, ProtLoc-GRPO, to enhance subcellular localization prediction by optimizing the structure of the underlying PPI network. ProtLoc-GRPO learns to rank and retain the most informative PPI edges to maximize the macro-F1 score for cell line-specific subcellular localization. Our approach yields a 7% improvement in macro-F1 score over the baseline. We further evaluate its robustness across various edge pruning rates and benchmark it against conventional pruning strategies. Results show that our proposed method consistently outperforms existing approaches. To our knowledge, this work represents the first sequence-based study to predict cell line-specific protein subcellular localization and the first application of the Group Relative Policy Optimization (GRPO) framework to a graph-based model for bioinformatics tasks.

Protein language models (PLMs) trained solely on sequence data have significantly advanced our understanding of protein biology and achieved remarkable performance in protein prediction tasks. However, their lack of three-dimensional (3D) structural features limits their predictive power in applications that rely heavily on 3D conformation. To address this limitation, we developed two structure-aware PLMs, S-PLM1 and S-PLM2, that employ multi-view contrastive learning to align protein sequences with their 3D structures in a unified latent space. S-PLM1 represents structural information using contact maps encoded by a pretrained Swin-Transformer, while S-PLM2 directly encodes 3D backbone coordinates through a Geometric Vector Perceptron (GVP)-based model. The paired sequence-structure data were obtained from AlphaFoldDB. For both models, we designed efficient tuning strategies that enable optimal performance with minimal computational cost. Here, we present detailed protocols for adapting S-PLM1 and S-PLM2 for diverse protein applications. The protocols provide step-by-step guidance on generating structure-aware representations from S-PLMs, fine-tuning them for various protein prediction tasks, and using S-PLM2 to produce structure embeddings for structure-based downstream analyses. We also provide source code and Google Colab implementations for easy customization and deployment. © 2026 Wiley Periodicals LLC. Basic Protocol 1: Generating structure-aware representations of protein sequences Basic Protocol 2: Efficient tuning of structure-aware protein language models for diverse protein applications Basic Protocol 3: Using S-PLM2 to generate protein structure representations and conduct structure-based clustering Support Protocol: Google Colab quick start notebooks.

Protein dynamics play a critical role in protein function, yet such important information is missing in many protein language models (PLM). We introduce DPLM, a dynamics-aware protein language model that aligns sequence embeddings with molecular dynamics (MD) trajectory embeddings via contrastive learning. Using MD features encoded by a pretrained video model, DPLM learns sequence representations that correlate with residue-level flexibility and improve protein-level functional clustering compared to static sequence- and structure-based PLMs. Without task-specific training, DPLM outperforms ESM-based representations in zero-shot mutation-effect prediction on multiple deep mutational scanning datasets. When adapted with lightweight task-specific heads, DPLM further achieves top-tier performance on protein stability prediction and intrinsic disorder region identification, demonstrating that contrastive alignment with MD trajectories enables PLMs to capture biologically meaningful dynamic properties.

Kinase-specific phosphorylation plays a critical role in cellular signaling and various diseases. However, even in model organisms, the substrates of most kinases remain unidentified. Currently, there is no reliable method to predict kinase-substrate relationships. In this study, we introduce an innovative approach leveraging an autoregressive model to predict kinase-substrate pairs. Unlike traditional methods focused on predicting site-specific phosphorylation, our approach addresses kinase-specific protein substrate prediction at the protein level. We redefine this problem as a special type of protein-protein interaction prediction task. Our model integrates protein large language model ESM-2 as the encoder and employs an autoregressive decoder to classify protein-kinase interactions in a binary fashion. We adopted a hard negative strategy, based on kinase embedding distances generated from ESM-2, to compel the model to effectively distinguish positive from negative data. We conducted a top‑k analysis to assess how well our model can prioritize the most likely kinase candidates. Our method is also capable of zero-shot prediction, meaning it can predict substrates for a kinase in case of no known substrates, which cannot be achieved by site-specific prediction methods. Our model’s robust generalization to novel kinase and underrepresented groups showcases its versatility and broad utility. Code and data are available at https://github.com/farz1995/substrate_kinase_prediction.

The diverse nature of protein prediction tasks has traditionally necessitated specialized models, hindering the development of broadly applicable and computationally efficient Protein Language Models (PLMs). In this work, we introduce Prot2Token, a unified framework that overcomes these challenges by converting a wide spectrum of protein-related predictions-from sequence-level properties and residue-specific attributes to complex inter-protein interactions-into a standardized next-token prediction format. At its core, Prot2Token employs an autoregressive decoder, conditioned on embeddings from pre-trained protein encoders and guided by learnable task tokens, to perform diverse predictions. This architecture uniquely facilitates multi-task learning, enabling general-purpose decoders to generalize across five distinct categories. We present extensive experimental validation across a variety of benchmarks, demonstrating Prot2Token's predictive power in different types of protein-prediction tasks. In 3D structure prediction, Prot2Token delivers substantial speedups (up to 1000x faster than AlphaFold2 with MSA on the same hardware) while, across other numerous tasks, matching or surpassing specialized methods. Beyond that, we introduce an auxiliary self-supervised decoder pre-training approach to improve spatially sensitive task performance. Prot2Token thus offers a step towards standardizing biological prediction into a generative interface, promising to accelerate biological discovery and the development of novel therapeutics. The code is available at https://github.com/mahdip72/prot2token .

Post-translational modifications (PTMs) regulate protein function, stability, and interactions, playing essential roles in cellular signaling, localization, and disease mechanisms. Computational approaches enable scalable PTM site prediction; however, traditional models focus only on local sequence features from fragments around potential modification sites, limiting the scope of their predictions. Recently, pre-trained protein language models (PLMs) have improved PTM prediction by leveraging biological knowledge derived from extensive protein databases. However, most PLMs used for PTM site prediction are pre-trained solely on amino acid sequences, limiting their ability to capture the structural context necessary for accurate PTM site prediction. Moreover, these methods typically train separate single-task models for each PTM type, which hinders the sharing of common features and limits potential knowledge transfer across tasks. To overcome these limitations, we introduce MTPrompt-PTM, a multi-task PTM prediction framework developed by applying prompt tuning to a structure-aware protein language model (S-PLM). Instead of training several single-task models, MTPrompt-PTM trains one multi-task model to predict multiple types of PTM sites using shared feature extraction layers and task-specific classification heads. Additionally, we incorporate a knowledge distillation strategy to enhance the efficiency and generalizability of multi-task training. Experimental results demonstrate that MTPrompt-PTM outperforms state-of-the-art PTM prediction tools on 13 types of PTM sites, highlighting the advantages of multi-task learning and structural integration.

Proteins are crucial in a wide range of biological and engineering processes. Large protein language models (PLMs) can significantly advance our understanding and engineering of proteins. However, the effectiveness of PLMs in prediction and design is largely based on the representations derived from protein sequences. Without incorporating the three-dimensional (3D) structures of proteins, PLMs would overlook crucial aspects of how proteins interact with other molecules, thereby limiting their predictive accuracy. To address this issue, we present S-PLM, a 3D structure-aware PLM, that employs multi-view contrastive learning to align protein sequences with their 3D structures in a unified latent space. Previously, we utilized a contact map-based approach to encode structural information, applying the Swin-Transformer to contact maps derived from AlphaFold-predicted protein structures. This work introduces a new approach that leverages a geometric vector perceptron (GVP) model to process 3D coordinates and obtain structural embeddings. We focus on the application of structure-aware models for protein-related tasks by utilizing efficient fine-tuning methods to achieve optimal performance without significant computational costs. Our results show that S-PLM outperforms sequence-only PLMs across all protein clustering and classification tasks, achieving performance on par with state-of-the-art methods that require both sequence and structure inputs. S-PLM and its tuning tools are available at https://github.com/duolinwang/S-PLM/ .

The immunogenic binding interactions of antigens are complex and interconnected. A new transformer-based model can simultaneously predict the bindings of antigens to two main receptors.

Protein targeting, often guided by targeting peptides, is a critical biological process that directs proteins to their specific cellular destinations, ensuring proper cellular functionality and organization. Accurate classification and detection of targeting peptides are fundamental to understanding protein sorting mechanisms. This study introduces MULoc-Target, a novel deep-learning method designed to detect and classify targeting peptides in eukaryotic proteins. To support its development and evaluation, we curated a benchmark dataset comprising eight types of eukaryotic targeting peptides with manually curated annotations. Comprehensive evaluations on this dataset and external datasets from the literature demonstrate that MULoc-Target achieves state-of-the-art or competitive performance in detecting and classifying targeting peptides. Additionally, it enables the extraction of enriched motif patterns, offering valuable insights into their properties and the underlying targeting mechanisms. The identified motifs align closely with established biological features, further validating MULoc-Target's capabilities. A web server for MULoc-Target is integrated into our MULocDeep localization suite as a new toolkit, publicly accessible at https://mu-loc.org/MULoc-Target, and the inference code is available at https://github.com/yuexujiang/MULoc-Target.

Peptide design, with the goal of identifying peptides possessing unique biological properties, stands as a crucial challenge in peptide-based drug discovery. While traditional and computational methods have made significant strides, they often encounter hurdles due to the complexities and costs of laboratory experiments. Recent advancements in deep learning and Bayesian Optimization have paved the way for innovative research in this domain. In this context, our study presents a novel approach that effectively combines protein structure prediction with Bayesian Optimization for peptide design. By applying carefully designed objective functions, we guide and enhance the optimization trajectory for new peptide sequences. Benchmarked against multiple native structures, our methodology is tailored to generate new peptides to their optimal potential biological properties.

This paper proposes a versatile tokenization method and introduces Prot2Token, a model that combines autoregressive language modeling with protein language models (PLMs) to tackle various protein prediction tasks using protein sequences. Leveraging our tokenization method, Prot2Token adapts existing PLMs for multiple tasks such as protein-level prediction, residue-level prediction, and protein-protein interaction prediction through next-token prediction of tokenized target label sequences. By incorporating prompt tokens into the decoder, Prot2Token enables multi-task training in a single end-to-end session. Our results demonstrate that Prot2Token not only matches the performance of specialized models across various tasks but also paves the way for integrating protein tasks with large language models (LLMs), representing an important step towards creating general-purpose PLMs for advanced protein language processing (PLP). Additionally, we use Prot2Token to develop S-ESM, a structure-aware version of the ESM model, which achieves competitive performance with state-of-the-art methods in 3D structure-related tasks using only protein sequences. Code is available at: . ### Competing Interest Statement The authors have declared no competing interest.

The DescribePROT database of amino acid-level descriptors of protein structures and functions was substantially expanded since its release in 2020. This expansion includes substantial increase in the size, scope, and quality of the underlying data, the addition of experimental structural information, the inclusion of new data download options, and an upgraded graphical interface. DescribePROT currently covers 19 structural and functional descriptors for proteins in 273 reference proteomes generated by 11 accurate and complementary predictive tools. Users can search our resource in multiple ways, interact with the data using the graphical interface, and download data at various scales including individual proteins, entire proteomes, and whole database. The annotations in DescribePROT are useful for a broad spectrum of studies that include investigations of protein structure and function, development and validation of predictive tools, and to support efforts in understanding molecular underpinnings of diseases and development of therapeutics. DescribePROT can be freely accessed at http://biomine.cs.vcu.edu/servers/DESCRIBEPROT/.

Signal peptides (SPs) play a crucial role in protein translocation in cells. The development of large protein language models (PLMs) and prompt-based learning provide a new opportunity for SP prediction, especially for the categories with limited annotated data. We present a parameter-efficient fine-tuning (PEFT) framework for SP prediction, PEFT-SP, to effectively utilize pretrained PLMs. We integrated low-rank adaptation (LoRA) into ESM-2 models to better leverage the protein sequence evolutionary knowledge of PLMs. Experiments show that PEFT-SP using LoRA enhances state-of-the-art results, leading to a maximum Matthews correlation coefficient (MCC) gain of 87.3% for SPs with small training samples and an overall MCC gain of 6.1%. Furthermore, we also employed two other PEFT methods, prompt tuning and adapter tuning, in ESM-2 for SP prediction. More elaborate experiments show that PEFT-SP using adapter tuning can also improve the state-of-the-art results by up to 28.1% MCC gain for SPs with small training samples and an overall MCC gain of 3.8%. LoRA requires fewer computing resources and less memory than the adapter tuning during the training stage, making it possible to adapt larger and more powerful protein models for SP prediction.

Pre-trained Language Models (PLMs) have demonstrated remarkable performance in Natural Language Understanding (NLU) tasks, with continuous prompt-based fine-tuning further enhancing their capabilities. However, current methods rely on hand-crafted discrete prompts to initialize continuous prompts, which are sensitive to subtle changes and inherently limited by the constraints of natural language. To address these limitations, this study introduces an innovative AutoPrompt-based Prompt Tuning (APT) approach. APT optimizes the initialization of continuous prompts by employing a gradient-guided automatic search to generate ideal discrete templates and identify trigger tokens. As the semantic features are already captured from the target task dataset, the continuous parameters initialized by trigger tokens are highly relevant, providing a superior starting point for prompt-tuning. APT searches for optimal prompts across various NLU tasks, enabling the PLM to learn task-related knowledge effectively. The APT method significantly improves PLM performance in both few-shot and fully supervised settings, eliminating the need for extensive prompt engineering. In the knowledge exploration (Language Model Analysis (LAMA)) benchmark, APT achieved a remarkable 58.6% (P@1) performance without additional text, representing a 3.6% improvement over the previous best result. Additionally, APT outperformed state-of-the-art methods in the SuperGLUE benchmark.

Introduction Immune checkpoint inhibitors (ICIs) are potent and precise therapies for various cancer types, significantly improving survival rates in patients who respond positively to them. However, only a minority of patients benefit from ICI treatments. Objectives Identifying ICI responders before treatment could greatly conserve medical resources, minimize potential drug side effects, and expedite the search for alternative therapies. Our goal is to introduce a novel deep-learning method to predict ICI treatment responses in cancer patients. Methods The proposed deep-learning framework leverages graph neural network and biological pathway knowledge. We trained and tested our method using ICI-treated patients’ data from several clinical trials covering melanoma, gastric cancer, and bladder cancer. Results Our results demonstrate that this predictive model outperforms current state-of-the-art methods and tumor microenvironment-based predictors. Additionally, the model quantifies the importance of pathways, pathway interactions, and genes in its predictions. A web server for IRnet has been developed and deployed, providing broad accessibility to users at https://irnet.missouri.edu. Conclusion IRnet is a competitive tool for predicting patient responses to immunotherapy, specifically ICIs. Its interpretability also offers valuable insights into the mechanisms underlying ICI treatments.

Denoising diffusion models embody a type of generative artificial intelligence that can be applied in computer vision, natural language processing and bioinformatics. In this Review, we introduce the key concepts and theoretical foundations of three diffusion modelling frameworks (denoising diffusion probabilistic models, noise-conditioned scoring networks and score stochastic differential equations). We then explore their applications in bioinformatics and computational biology, including protein design and generation, drug and small-molecule design, protein–ligand interaction modelling, cryo-electron microscopy image data analysis and single-cell data analysis. Finally, we highlight open-source diffusion model tools and consider the future applications of diffusion models in bioinformatics. Diffusion models are deep-learning-based generative models that can generate new data from input parameters. This Review discusses applications of diffusion models in bioinformatics and computational biology.

Proteins play an essential role in various biological and engineering processes. Large protein language models (PLMs) present excellent potential to reshape protein research by accelerating the determination of protein function and the design of proteins with the desired functions. The prediction and design capacity of PLMs relies on the representation gained from the protein sequences. However, the lack of crucial 3D structure information in most PLMs restricts the prediction capacity of PLMs in various applications, especially those heavily dependent on 3D structures. To address this issue, we introduce S-PLM, a 3D structure-aware PLM that utilizes multi-view contrastive learning to align the sequence and 3D structure of a protein in a coordinated latent space. S-PLM applies Swin-Transformer on AlphaFold-predicted protein structures to embed the structural information and fuses it into sequence-based embedding from ESM2. Additionally, we provide a library of lightweight tuning tools to adapt S-PLM for diverse protein property prediction tasks. Our results demonstrate S-PLM's superior performance over sequence-only PLMs on all protein clustering and classification tasks, achieving competitiveness comparable to state-of-the-art methods requiring both sequence and structure inputs. S-PLM and its lightweight tuning tools are available at https://github.com/duolinwang/S-PLM/ .

Antibiofouling peptide materials prevent the nonspecific adsorption of proteins on devices, enabling them to perform their designed functions as desired in complex biological environments. Due to their importance, research on antibiofouling peptide materials has been one of the central subjects of interfacial engineering. However, only a few antibiofouling peptide sequences have been developed. This narrow scope of antibiofouling peptide materials limits their capacity to adapt to the broad spectrum of application scenarios. To address this issue, we searched for antibiofouling peptides in the vast sequence pool of the microbiome library using a combination of deep learning-based high-throughput search and molecular dynamics (MD) simulations. A random forest-based model with an ensemble of ten independent classifiers was developed. Each classifier was trained by prompt-tuning the foundational protein language model Evolution Scaling Modeling version 2 (ESM2) on a distinct training data set. We constructed the databases containing the same amount of antibiofouling and biofouling peptide sequences to attenuate the bias of the existing databases. MD simulations were conducted to investigate the interfacial properties of six selected peptide candidates and their interactions with a lysozyme protein. Two known antibiofouling peptides, (glutamic acid (E)-lysine (K))15 and (EK-proline (P))10, and one known fouling peptide, (glycine)30, were used as the reference. The MD simulation results indicate that five of the six peptides present the potential to resist biofouling. Our research implies that deep learning and molecular simulations can be integrated to discover functional peptide materials for interfacial applications.

Signal peptides (SP) play a crucial role in protein localization in cells. The development of large protein language models (PLMs) provides a new opportunity for SP prediction. We applied a prompt-based learning framework, Parameter-Efficient Fine-Tuning (PEFT) for SP prediction, PEFT-SP, to effectively utilize pre-trained PLMs. We integrated low-rank adaptation (LoRA) into ESM-2 models to better leverage the protein sequence evolutionary knowledge of PLMs. Experiments show that PEFT-SP using LoRA enhances state-of-the-art results, leading to a maximum MCC gain of 0.372 for SPs with small training samples and an overall MCC gain of 0.048. Furthermore, we also employed two other prompt-based learning methods, i.e., Prompt Tuning and Adapter Tuning, into ESM-2 for SP prediction. More elaborate experiments show that PEFT-SP using Adapter Tuning can also improve the state-of-the-art results with up to 0.202 MCC gain for SPs with small training samples and an overall MCC gain of 0.030. LoRA requires fewer computing resources and less memory than the Adapter during the training stage, making it possible to adapt larger and more powerful protein models for SP prediction. The PEFT-SP framework is available at https://github.com/shuaizengMU/PEFT-SP. The web server for SP predic-tion leveraging the PEFT-SP framework is publicly available at https://www.mu-loc.org/peftsp/.