We introduce Ibex, a pan-immunoglobulin structure prediction model that achieves state-of-the-art accuracy in modeling the variable domains of antibodies, nanobodies, and T-cell receptors. Unlike previous approaches, Ibex explicitly distinguishes between bound and unbound protein conformations by training on labeled apo and holo structural pairs, enabling accurate prediction of both states at inference time. Using a comprehensive private dataset of high-resolution antibody structures, we demonstrate superior out-of-distribution performance compared to existing specialized and general protein structure prediction tools. Ibex combines the accuracy of cutting-edge models with significantly reduced computational requirements, providing a robust foundation for accelerating large molecule design and therapeutic development.
Protein structure prediction models such as AlphaFold3 (AF3) push the frontier of biomolecular modeling by incorporating science-informed architectural changes to the transformer architecture. However, these advances come at a steep system cost, introducing: compute- and memory-intensive operators, 2D attention mechanisms, and retrieval-augmented data pipelines, which collectively hinder the scalability of AF3 training. In this work, we present MegaFold, a cross-platform system to accelerate AF3 training. MegaFold tackles key bottlenecks through ahead-of-time caching to eliminate GPU idle time from the retrieval-augmented data pipeline, Triton-based kernels for memory-efficient EvoAttention on heterogeneous devices, and deep fusion for common and critical small operators in AF3. Evaluation on both NVIDIA H200 and AMD MI250 GPUs shows that MegaFold reduces peak memory usage of AF3 training by up to 1.23$\times$ and improves per-iteration training time by up-to 1.73$\times$ and 1.62$\times$ respectively. More importantly, MegaFold enables training on 1.35$\times$ longer sequence lengths compared to PyTorch baselines without running out-of-memory, significantly improving the scalability of modern protein folding models. We open source our code at https://github.com/Supercomputing-System-AI-Lab/MegaFold/.
Breakthroughs in high-accuracy protein structure prediction, such as AlphaFold, have established receptor-based molecule design as a critical driver for rapid early-phase drug discovery. However, most approaches still struggle to balance pocket-specific geometric fit with strict valence and synthetic constraints. To resolve this trade-off, a Retrieval-Enhanced Aligned Diffusion termed READ is introduced, which is the first to merge molecular Retrieval-Augmented Generation with an SE(3)-equivariant diffusion model. Specifically, a contrastively pre-trained encoder aligns atom-level representations during training, then retrieves graph embeddings of pocket-matched scaffolds to guide each reverse-diffusion step at inference. This single mechanism can inject real-world chemical priors exactly where needed, producing valid, diverse, and shape-complementary ligands. Experimental results demonstrate that READ can achieve very competitive performance in CBGBench, surpassing state-of-the-art generative models and even native ligands. That suggests retrieval and diffusion can be co-optimized for faster, more reliable structure-based drug design.
Protein structure prediction models are now capable of generating accurate 3D structural hypotheses from sequence alone. However, they routinely fail to capture the conformational diversity of dynamic biomolecular complexes, often requiring heuristic MSA subsampling approaches for generating alternative states. In parallel, cryo-electron microscopy (cryo-EM) has emerged as a powerful tool for imaging near-native structural heterogeneity, but is challenged by arduous pipelines to go from raw experimental data to atomic models. Here, we bridge the gap between these modalities, combining cryo-EM density maps with the rich sequence and biophysical priors learned by protein structure prediction models. Our method, CryoBoltz, guides the sampling trajectory of a pretrained protein structure prediction model using both global and local structural constraints derived from density maps, driving predictions towards conformational states consistent with the experimental data. We demonstrate that this flexible yet powerful inference-time approach allows us to build atomic models into heterogeneous cryo-EM maps across a variety of dynamic biomolecular systems including transporters and antibodies.
Protein dynamics play a crucial role in protein biological functions and properties, and their traditional study typically relies on time-consuming molecular dynamics (MD) simulations conducted in silico. Recent advances in generative modeling, particularly denoising diffusion models, have enabled efficient accurate protein structure prediction and conformation sampling by learning distributions over crystallographic structures. However, effectively integrating physical supervision into these data-driven approaches remains challenging, as standard energy-based objectives often lead to intractable optimization. In this paper, we introduce Energy-based Alignment (EBA), a method that aligns generative models with feedback from physical models, efficiently calibrating them to appropriately balance conformational states based on their energy differences. Experimental results on the MD ensemble benchmark demonstrate that EBA achieves state-of-the-art performance in generating high-quality protein ensembles. By improving the physical plausibility of generated structures, our approach enhances model predictions and holds promise for applications in structural biology and drug discovery.
Predicting protein complex structures is essential for protein function analysis, protein design, and drug discovery. While AI methods like AlphaFold can predict accurate structural models for many protein complexes, reliably estimating the quality of these predicted models (estimation of model accuracy, or EMA) for model ranking and selection remains a major challenge. A key barrier to developing effective machine learning-based EMA methods is the lack of large, diverse, and well-annotated datasets for training and evaluation. To address this gap, we introduce PSBench, a benchmark suite comprising four large-scale, labeled datasets generated during the 15th and 16th community-wide Critical Assessment of Protein Structure Prediction (CASP15 and CASP16). PSBench includes over one million structural models covering a wide range of protein sequence lengths, complex stoichiometries, functional classes, and modeling difficulties. Each model is annotated with multiple complementary quality scores at the global, local, and interface levels. PSBench also provides multiple evaluation metrics and baseline EMA methods to facilitate rigorous comparisons. To demonstrate PSBench's utility, we trained and evaluated GATE, a graph transformer-based EMA method, on the CASP15 data. GATE was blindly tested in CASP16 (2024), where it ranked among the top-performing EMA methods. These results highlight PSBench as a valuable resource for advancing EMA research in protein complex modeling. PSBench is publicly available at: https://github.com/BioinfoMachineLearning/PSBench.
We present a novel theoretical interpretation of AlphaFold1. The seminal breakthrough of AlphaFold1 in protein structure prediction by deep learning relied on a learned potential energy function, in contrast to the later end-to-end architectures of AlphaFold2 and AlphaFold3. While this potential was originally justified by referring to physical potentials of mean force (PMFs), we reinterpret AlphaFold1's potential as an instance of probability kinematics - also known as Jeffrey conditioning - a principled but underrecognised generalization of conventional Bayesian updating. Probability kinematics accommodates uncertain or soft evidence in the form of updated probabilities over a partition. This perspective reveals AlphaFold1's potential as a form of generalized Bayesian updating, rather than a thermodynamic potential. To confirm our probabilistic framework's scope and precision, we analyze a synthetic 2D model in which an angular random walk prior is updated with evidence on distances via probability kinematics, mirroring AlphaFold1's approach. This theoretical contribution connects AlphaFold1 to a broader class of well-justified Bayesian methods, allowing precise quantification, surpassing merely qualitative heuristics based on PMFs. More broadly, given the achievements of AlphaFold1, probability kinematics holds considerable promise for probabilistic deep learning, as it allows for the formulation of complex models from a few simpler components.
As protein informatics advances rapidly, the demand for enhanced predictive accuracy, structural analysis, and functional understanding has intensified. Transformer models, as powerful deep learning architectures, have demonstrated unprecedented potential in addressing diverse challenges across protein research. However, a comprehensive review of Transformer applications in this field remains lacking. This paper bridges this gap by surveying over 100 studies, offering an in-depth analysis of practical implementations and research progress of Transformers in protein-related tasks. Our review systematically covers critical domains, including protein structure prediction, function prediction, protein-protein interaction analysis, functional annotation, and drug discovery/target identification. To contextualize these advancements across various protein domains, we adopt a domain-oriented classification system. We first introduce foundational concepts: the Transformer architecture and attention mechanisms, categorize Transformer variants tailored for protein science, and summarize essential protein knowledge. For each research domain, we outline its objectives and background, critically evaluate prior methods and their limitations, and highlight transformative contributions enabled by Transformer models. We also curate and summarize pivotal datasets and open-source code resources to facilitate reproducibility and benchmarking. Finally, we discuss persistent challenges in applying Transformers to protein informatics and propose future research directions. This review aims to provide a consolidated foundation for the synergistic integration of Transformer and protein informatics, fostering further innovation and expanded applications in the field.
Recent advances in Protein Structure Prediction Models (PPMs), such as AlphaFold2 and ESMFold, have revolutionized computational biology by achieving unprecedented accuracy in predicting three-dimensional protein folding structures. However, these models face significant scalability challenges, particularly when processing proteins with long amino acid sequences (e.g., sequence length > 1,000). The primary bottleneck that arises from the exponential growth in activation sizes is driven by the unique data structure in PPM, which introduces an additional dimension that leads to substantial memory and computational demands. These limitations have hindered the effective scaling of PPM for real-world applications, such as analyzing large proteins or complex multimers with critical biological and pharmaceutical relevance. In this paper, we present LightNobel, the first hardware-software co-designed accelerator developed to overcome scalability limitations on the sequence length in PPM. At the software level, we propose Token-wise Adaptive Activation Quantization (AAQ), which leverages unique token-wise characteristics, such as distogram patterns in PPM activations, to enable fine-grained quantization techniques without compromising accuracy. At the hardware level, LightNobel integrates the multi-precision reconfigurable matrix processing unit (RMPU) and versatile vector processing unit (VVPU) to enable the efficient execution of AAQ. Through these innovations, LightNobel achieves up to 8.44x, 8.41x speedup and 37.29x, 43.35x higher power efficiency over the latest NVIDIA A100 and H100 GPUs, respectively, while maintaining negligible accuracy loss. It also reduces the peak memory requirement up to 120.05x in PPM, enabling scalable processing for proteins with long sequences.
Protein structure prediction is a critical and longstanding challenge in biology, garnering widespread interest due to its significance in understanding biological processes. A particular area of focus is the prediction of missing loops in proteins, which are vital in determining protein function and activity. To address this challenge, we propose AutoLoop, a novel computational model designed to automatically generate accurate loop backbone conformations that closely resemble their natural structures. AutoLoop employs a bidirectional training approach while merging atom- and residue-level embedding, thus improving robustness and precision. We compared AutoLoop with twelve established methods, including FREAD, NGK, AlphaFold2, and AlphaFold3. AutoLoop consistently outperforms other methods, achieving a median RMSD of 1.12 Angstrom and a 2-Angstrom success rate of 73.23% on the CASP15 dataset, while maintaining strong performance on the HOMSTARD dataset. It demonstrates the best performance across nearly all loop lengths and secondary structural types. Beyond accuracy, AutoLoop is computationally efficient, requiring only 0.10 s per generation. A post-processing module for side-chain packing and energy minimization further improves results slightly, confirming the reliability of the predicted backbone. A case study also highlights AutoLoop's potential for precise predictions based on dominant loop conformations. These advances hold promise for protein engineering and drug discovery.
The ability to make zero-shot predictions about the fitness consequences of protein sequence changes with pre-trained machine learning models enables many practical applications. Such models can be applied for downstream tasks like genetic variant interpretation and protein engineering without additional labeled data. The advent of capable protein structure prediction tools has led to the availability of orders of magnitude more precomputed predicted structures, giving rise to powerful structure-based fitness prediction models. Through our experiments, we assess several modeling choices for structure-based models and their effects on downstream fitness prediction. Zero-shot fitness prediction models can struggle to assess the fitness landscape within disordered regions of proteins, those that lack a fixed 3D structure. We confirm the importance of matching protein structures to fitness assays and find that predicted structures for disordered regions can be misleading and affect predictive performance. Lastly, we evaluate an additional structure-based model on the ProteinGym substitution benchmark and show that simple multi-modal ensembles are strong baselines.
The 2024 Nobel Prize in Chemistry was awarded in part for protein structure prediction using AlphaFold2, an artificial intelligence/machine learning (AI/ML) model trained on vast amounts of sequence and 3D structure data. AlphaFold2 and related models, including RoseTTAFold and ESMFold, employ specialized neural network architectures driven by attention mechanisms to infer relationships between sequence and structure. At a fundamental level, these AI/ML models operate on the long-standing hypothesis that the structure of a protein is determined by its amino acid sequence. More recently, AlphaFold2 has been adapted for the prediction of multiple protein conformations by subsampling multiple sequence alignments (MSAs). The deterministic relationship between sequence and structure was hypothesized over half a century ago with profound implications for the biological sciences ever since. Based on this relationship, we hypothesize that protein conformational dynamics are also determined, at least in part, by amino acid sequence and that this relationship may be leveraged for construction of AI/ML models dedicated to predicting ensembles of protein structures (i.e., distinct conformations). Accordingly, we conceptualized an AI/ML model architecture which may be trained on sequence data in combination with conformationally-sensitive structure data, coming primarily from nuclear magnetic resonance (NMR) spectroscopy. Sequence-informed prediction of protein structural dynamics has the potential to emerge as a transformative capability across the biological sciences, and its implementation could very well be on the horizon.
Scientific knowledge creation is fundamentally transforming as humans and AI systems evolve beyond tool-user relationships into co-evolutionary epistemic partnerships. When AlphaFold revolutionized protein structure prediction, researchers described engaging with an epistemic partner that reshaped how they conceptualized fundamental relationships. This article introduces Cognitio Emergens (CE), a framework addressing critical limitations in existing models that focus on static roles or narrow metrics while failing to capture how scientific understanding emerges through recursive human-AI interaction over time. CE integrates three components addressing these limitations: Agency Configurations describing how authority distributes between humans and AI (Directed, Contributory, Partnership), with partnerships dynamically oscillating between configurations rather than following linear progression; Epistemic Dimensions capturing six specific capabilities emerging through collaboration across Discovery, Integration, and Projection axes, creating distinctive "capability signatures" that guide development; and Partnership Dynamics identifying forces shaping how these relationships evolve, particularly the risk of epistemic alienation where researchers lose interpretive control over knowledge they formally endorse. Drawing from autopoiesis theory, social systems theory, and organizational modularity, CE reveals how knowledge co-creation emerges through continuous negotiation of roles, values, and organizational structures. By reconceptualizing human-AI scientific collaboration as fundamentally co-evolutionary, CE offers a balanced perspective that neither uncritically celebrates nor unnecessarily fears AI's evolving role, instead providing conceptual tools for cultivating partnerships that maintain meaningful human participation while enabling transformative scientific breakthroughs.
Antibiotic resistance presents a growing global health crisis, demanding new therapeutic strategies that target novel bacterial mechanisms. Recent advances in protein structure prediction and machine learning-driven molecule generation offer a promising opportunity to accelerate drug discovery. However, practical guidance on selecting and integrating these models into real-world pipelines remains limited. In this study, we develop an end-to-end, artificial intelligence-guided antibiotic discovery pipeline that spans target identification to compound realization. We leverage structure-based clustering across predicted proteomes of multiple pathogens to identify conserved, essential, and non-human-homologous targets. We then systematically evaluate six leading 3D-structure-aware generative models$\unicode{x2014}$spanning diffusion, autoregressive, graph neural network, and language model architectures$\unicode{x2014}$on their usability, chemical validity, and biological relevance. Rigorous post-processing filters and commercial analogue searches reduce over 100 000 generated compounds to a focused, synthesizable set. Our results highlight DeepBlock and TamGen as top performers across diverse criteria, while also revealing critical trade-offs between model complexity, usability, and output quality. This work provides a comparative benchmark and blueprint for deploying artificial intelligence in early-stage antibiotic development.
Proteins move and deform to ensure their biological functions. Despite significant progress in protein structure prediction, approximating conformational ensembles at physiological conditions remains a fundamental open problem. This paper presents a novel perspective on the problem by directly targeting continuous compact representations of protein motions inferred from sparse experimental observations. We develop a task-specific loss function enforcing data symmetries, including scaling and permutation operations. Our method PETIMOT (Protein sEquence and sTructure-based Inference of MOTions) leverages transfer learning from pre-trained protein language models through an SE(3)-equivariant graph neural network. When trained and evaluated on the Protein Data Bank, PETIMOT shows superior performance in time and accuracy, capturing protein dynamics, particularly large/slow conformational changes, compared to state-of-the-art flow-matching approaches and traditional physics-based models.
Evolution-based protein structure prediction models have achieved breakthrough success in recent years. However, they struggle to generalize beyond evolutionary priors and on sequences lacking rich homologous data. Here we present a novel, out-of-domain benchmark based on sactipeptides, a rare class of ribosomally synthesized and post-translationally modified peptides (RiPPs) characterized by sulfur-to-$\alpha$-carbon thioether bridges creating cross-links between cysteine residues and backbone. We evaluate recent models on predicting conformations compatible with these cross-links bridges for the 10 known sactipeptides with elucidated post-translational modifications. Crucially, the structures of 5 of them have not yet been experimentally resolved. This makes the task a challenging problem for evolution-based models, which we find exhibit limited performance (0.0% to 19.2% GDT-TS on sulfur-to-$\alpha$-carbon distance). Our results point at the need for physics-informed models to sustain progress in biomolecular structure prediction.
After AlphaFold won the Nobel Prize, protein prediction with deep learning once again became a hot topic. We comprehensively explore advanced deep learning methods applied to protein structure prediction and design. It begins by examining recent innovations in prediction architectures, with detailed discussions on improvements such as diffusion based frameworks and novel pairwise attention modules. The text analyses key components including structure generation, evaluation metrics, multiple sequence alignment processing, and network architecture, thereby illustrating the current state of the art in computational protein modelling. Subsequent chapters focus on practical applications, presenting case studies that range from individual protein predictions to complex biomolecular interactions. Strategies for enhancing prediction accuracy and integrating deep learning techniques with experimental validation are thoroughly explored. The later sections review the industry landscape of protein design, highlighting the transformative role of artificial intelligence in biotechnology and discussing emerging market trends and future challenges. Supplementary appendices provide essential resources such as databases and open source tools, making this volume a valuable reference for researchers and students.
Protein language models have revolutionized structure prediction, but their nonlinear nature obscures how sequence representations inform structure prediction. While sparse autoencoders (SAEs) offer a path to interpretability here by learning linear representations in high-dimensional space, their application has been limited to smaller protein language models unable to perform structure prediction. In this work, we make two key advances: (1) we scale SAEs to ESM2-3B, the base model for ESMFold, enabling mechanistic interpretability of protein structure prediction for the first time, and (2) we adapt Matryoshka SAEs for protein language models, which learn hierarchically organized features by forcing nested groups of latents to reconstruct inputs independently. We demonstrate that our Matryoshka SAEs achieve comparable or better performance than standard architectures. Through comprehensive evaluations, we show that SAEs trained on ESM2-3B significantly outperform those trained on smaller models for both biological concept discovery and contact map prediction. Finally, we present an initial case study demonstrating how our approach enables targeted steering of ESMFold predictions, increasing structure solvent accessibility while fixing the input sequence. To facilitate further investigation by the broader community, we open-source our code, dataset, pretrained models https://github.com/johnyang101/reticular-sae , and visualizer https://sae.reticular.ai .
AlphaFold2 (AF2) has transformed protein structure prediction by harnessing co-evolutionary constraints embedded in multiple sequence alignments (MSAs). MSAs not only encode static structural information, but also hold critical details about protein dynamics, which underpin biological functions. However, these subtle co-evolutionary signatures, which dictate conformational state preferences, are often obscured by noise within MSA data and thus remain challenging to decipher. Here, we introduce AF-ClaSeq, a systematic framework that isolates these co-evolutionary signals through sequence purification and iterative enrichment. By extracting sequence subsets that preferentially encode distinct structural states, AF-ClaSeq enables high-confidence predictions of alternative conformations. Our findings reveal that the successful sampling of alternative states depends not on MSA depth but on sequence purity. Intriguingly, purified sequences encoding specific structural states are distributed across phylogenetic clades and superfamilies, rather than confined to specific lineages. Expanding upon AF2's transformative capabilities, AF-ClaSeq provides a powerful approach for uncovering hidden structural plasticity, advancing allosteric protein and drug design, and facilitating dynamics-based protein function annotation.
Deep learning has transformed protein design, enabling accurate structure prediction, sequence optimization, and de novo protein generation. Advances in single-chain protein structure prediction via AlphaFold2, RoseTTAFold, ESMFold, and others have achieved near-experimental accuracy, inspiring successive work extended to biomolecular complexes via AlphaFold Multimer, RoseTTAFold All-Atom, AlphaFold 3, Chai-1, Boltz-1 and others. Generative models such as ProtGPT2, ProteinMPNN, and RFdiffusion have enabled sequence and backbone design beyond natural evolution-based limitations. More recently, joint sequence-structure co-design models, including ESM3, have integrated both modalities into a unified framework, resulting in improved designability. Despite these advances, challenges still exist pertaining to modeling sequence-structure-function relationships and ensuring robust generalization beyond the regions of protein space spanned by the training data. Future advances will likely focus on joint sequence-structure-function co-design frameworks that are able to model the fitness landscape more effectively than models that treat these modalities independently. Current capabilities, coupled with the dizzying rate of progress, suggest that the field will soon enable rapid, rational design of proteins with tailored structures and functions that transcend the limitations imposed by natural evolution. In this review, we discuss the current capabilities of deep learning methods for protein design, focusing on some of the most revolutionary and capable models with respect to their functionality and the applications that they enable, leading up to the current challenges of the field and the optimal path forward.