Robust deep learning–based protein sequence design using ProteinMPNN

2022/09/15 by Justas Dauparas, Ivan Anishchenko, Nathaniel R. Bennett +19 · 108 citations
Biochemistry, Genetics and Molecular Biology · #Machine Learning in Bioinformatics #Protein Structure and Dynamics #RNA and protein synthesis mechanisms

paper · doi:10.1126/science.add2187

openalex publication_date 2022/09/15 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/31

Abstract

Although deep learning has revolutionized protein structure prediction, almost all experimentally characterized de novo protein designs have been generated using physically based approaches such as Rosetta. Here, we describe a deep learning-based protein sequence design method, ProteinMPNN, that has outstanding performance in both in silico and experimental tests. On native protein backbones, ProteinMPNN has a sequence recovery of 52.4% compared with 32.9% for Rosetta. The amino acid sequence at different positions can be coupled between single or multiple chains, enabling application to a wide range of current protein design challenges. We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo-electron microscopy, and functional studies by rescuing previously failed designs, which were made using Rosetta or AlphaFold, of protein monomers, cyclic homo-oligomers, tetrahedral nanoparticles, and target-binding proteins.

Citations

• <i>Phaser</i>crystallographic software
• Neural Message Passing for Quantum Chemistry
• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Learning from Protein Structure with Geometric Vector Perceptrons
• Highly accurate protein structure prediction with AlphaFold
• Accurate prediction of protein structures and interactions using a three-track neural network
• Learning inverse folding from millions of predicted structures

Cited by

• Improving Protein Expression, Stability, and Function with ProteinMPNN
• Artificial intelligence in bioinformatics: a survey
• De novo design of miniproteins targeting GPCRs
• How far can you go? Extrapolating values of catalytic activity from known protein landscapes in natural and directed evolution
• Using a GPT-5-driven autonomous lab to optimize the cost and titer of cell-free protein synthesis
• Generalizable and scalable protein stability prediction with rewired protein generative models
• RAG-ESM: Improving pretrained protein language models via sequence retrieval
• AI‐Physics‐Experiment Trinity for Integrated Protein Dynamics Modeling
• Generative AI for synthetic biology: Designing biological parts, circuits, and genomes
• Artificial intelligence-driven computational methods for antibody design and optimization
• AlphaDesign: a de novo protein design framework based on AlphaFold
• Variant effect predictor correlation with functional assays is reflective of clinical classification performance
• De novo design of protein structure and function with RFdiffusion
• From sequence to scaffold: Computational design of protein nanoparticle vaccines from AlphaFold2-predicted building blocks
• InstructPLM-mu: 1-Hour Fine-Tuning of ESM2 Beats ESM3 in Protein Mutation Predictions
• Evolutionary-scale prediction of atomic-level protein structure with a language model
• Limitations of the refolding pipeline for de novo protein design
• Acquiring Improved Protein Variants With Probabilistic Preferential Learning
• AutoScientists: Self-Organizing Agent Teams for Long-Running Scientific Experimentation
• EnzyControl: Adding Functional and Substrate-Specific Control for Enzyme Backbone Generation
• LDBT instead of DBTL: combining machine learning and rapid cell-free testing
• Learning the shape of protein microenvironments with a holographic convolutional neural network
• BioPipelines: Accessible Computational Protein and Ligand Design for Chemical Biologists
• Relaxed Sequence Sampling for Diverse Protein Design
• Structural Enzymology, Phylogenetics, Differentiation, and Symbolic Reflexivity at the Dawn of Biology
• ODesign: A World Model for Biomolecular Interaction Design
• TEMPO: Temporal Multi-scale Autoregressive Generation of Protein Conformational Ensembles
• RiboPO: Preference Optimization for Structure- and Stability-Aware RNA Design
• Protein generation with embedding learning for motif diversification
• Inference-Time Compute Scaling For Flow Matching
• Crystallization and initial X-ray crystallographic analysis of a <i>de novo</i> -designed protein with left-handed βαβ units
• Atomically accurate de novo design of antibodies with RFdiffusion
• Protein Language Models: Is Scaling Necessary?
• Prediction-Augmented Trees for Reliable Statistical Inference
• Kernel-Based Evaluation of Conditional Biological Sequence Models
• RNAGenScape: Property-guided Optimization and Interpolation of mRNA Sequences with Manifold Langevin Dynamics
• Flows, straight but not so fast: Exploring the design space of Rectified Flows in Protein Design
• PRISM: Enhancing Protein Inverse Folding through Fine-Grained Retrieval on Structure-Sequence Multimodal Representations
• Latent Retrieval Augmented Generation of Cross-Domain Protein Binders
• ProteinAE: Protein Diffusion Autoencoders for Structure Encoding
• Calibrating Generative Models
• Re‐engineering of a carotenoid‐binding protein based on <scp>NMR</scp> structure
• Tpgen: a language model for stable protein design with a specific topology structure
• The turn less taken: Investigating patterns in <i>β</i> -turn dynamics using large-scale molecular dynamics data
• Higher-order assembly of a type IX retron enables exploitation for designer antimicrobials
• Autonomous Agents for Scientific Discovery: Orchestrating Scientists, Language, Code, and Physics
• BioBlobs: Unsupervised Discovery of Functional Substructures for Protein Function Prediction
• Evolutionary Profiles for Protein Fitness Prediction
• Enzyme structure correlates with variant effect predictability
• Adaptive Protein Design Protocols and Middleware
• Physicochemically Informed Dual-Conditioned Generative Model of T-Cell Receptor Variable Regions for Cellular Therapy
• ATOM: A Pretrained Neural Operator for Multitask Molecular Dynamics
• Distilled Protein Backbone Generation
• NS-Pep: De novo Peptide Design with Non-Standard Amino Acids
• Flow Autoencoders are Effective Protein Tokenizers
• Let Physics Guide Your Protein Flows: Topology-aware Unfolding and Generation
• Is Sequence Information All You Need for Bayesian Optimization of Antibodies?
• Système de double hybride bactérien quantitatif optimisé (qB2H) pour l'évaluation des interactions protéine–protéine
• Protein language models trained on biophysical dynamics inform mutation effects
• Discontinuous Epitope Fragments as Sufficient Target Templates for Efficient Binder Design
• Defining latent spaces by example: optimisation over the outputs of generative models
• Rational Design Principles for <i>De Novo</i> α-Helical Peptide Barrels with Dynamic Conductive Channels
• Computational Design of Peptide Assemblies
• Context-aware geometric deep learning for protein sequence design
• Scalable protein design using optimization in a relaxed sequence space
• Protein design drives synthetic biology research of plant natural products
• SwitchCraft: A Programmatic Framework for Designing State-Switching Proteins
• Building an expanded bio-based economy through synthetic biology
• <i>De Novo</i> Design of a Metalloprotein with a Synthetically Inspired Dinuclear Paddlewheel Coordination Motif
• RoseTTAFold diffusion-guided short peptide design: a case study of binders against Keap1/Nrf2
• Rise of the RNA machines – self-amplification in mRNA vaccine design
• Training Compute-Optimal Protein Language Models
• QOBRA: A Quantum Operator-Based Autoencoder for <i>De Novo</i> Molecular Design
• AI-assisted protein design to rapidly convert antibody sequences to intrabodies targeting diverse peptides and histone modifications
• PERAN INTUISI DALAM PROSES MERANCANG
• Protein design accelerates the development and application of optogenetic tools
• Synthetic biology and artificial intelligence in crop improvement
• A Perspective on Artificial Intelligence for Molecular Pathologists
• Molecular Display of the Animal Meta-Venome for Discovery of Novel Therapeutic Peptides
• Contextual protein and antibody encodings from equivariant graph transformers
• Tools For Building Artificial Biological Nanostructures
• Evaluating zero‐shot prediction of monomeric protein design success by <scp>AlphaFold</scp> , <scp>ESMFold</scp> , and <scp>ProteinMPNN</scp>
• De novo protein design: a transformative frontier in clinical protein applications
• Protein engineering in the deep learning era
• Discovery, design, and engineering of enzymes based on molecular retrobiosynthesis
• High-Affinity Protein Binder Design via Flow Matching and In Silico Maturation
• RemoteFoldSet: Benchmarking Structural Awareness of Protein Language Models
• Guided Sequence-Structure Generative Modeling for Iterative Antibody Optimization
• Monte Carlo Tree Diffusion with Multiple Experts for Protein Design
• <i>De novo</i> Design of All-atom Biomolecular Interactions with RFdiffusion3
• Harnessing the power of artificial intelligence to advance cell therapy
• Tokenizing Loops of Antibodies
• Scalable Coupling of Deep Learning with Logical Reasoning
• Topotein: Topological Deep Learning for Protein Representation Learning
• Computational design of a soluble mimic of the outer membrane <scp>LPS</scp> transport protein <scp>LptD</scp> suitable for screening of antibiotics
• Scaling Neuro-symbolic Problem Solving: Solver-Free Learning of Constraints and Objectives
• Supervised learning of protein variant effects across large‐scale mutagenesis datasets
• Energy-Based Flow Matching for Generating 3D Molecular Structure
• De novo protein design enables targeting of intractable oncogenic interfaces
• FrustraMPNN: An ultra-fast deep learning tool for proteome-scale analysis of deep mutational single-residue local energetic frustration in proteins
• AlphaFold reveals but sometimes distorts an organizational principle of protein folding
• Designing de novo TIM Barrels: Insights into Stabilization, Diversification, and Functionalization Strategies
• Decoding Polyphenol-Protein Interactions with Deep Learning: From Molecular Mechanisms to Food Applications
• Multi-state Protein Design with DynamicMPNN
• Latent-X: An Atom-level Frontier Model for De Novo Protein Binder Design
• Learning from B Cell Evolution: Adaptive Multi-Expert Diffusion for Antibody Design via Online Optimization
• ThermoRL:Structure-Aware Reinforcement Learning for Protein Mutation Design to Enhance Thermostability
• Demystify Protein Generation with Hierarchical Conditional Diffusion Models
• Ancestral Sequence Reconstruction to Accelerate Non-heme Iron-dependent Biocatalyst Engineering

Related

• A Model-Centric Review of Deep Learning for Protein Design
• Designing novel protein structures using sequence generator and AlphaFold2
• A Survey of Deep Learning Methods in Protein Bioinformatics and its Impact on Protein Design
• Deep Learning-Driven Protein Structure Prediction and Design: Key Model Developments by Nobel Laureates and Multi-Domain Applications
• Leveraging Deep Generative Model For Computational Protein Design And Optimization