Geometric deep learning assists protein engineering. Opportunities and Challenges

2025/06/19 by García-Vinuesa, Julián, Rojas, Jorge, Soto-García, Nicole +7 · 1 citation
#FOS: Biological sciences #Quantitative Methods (q-bio.QM)

paper · doi:10.48550/arxiv.2506.16091

Abstract

Protein engineering is experiencing a paradigmatic shift through the integration of geometric deep learning into computational design workflows. While traditional strategies, such as rational design and directed evolution, have enabled relevant advances, they remain limited by the complexity of sequence space and the cost of experimental validation. Geometric deep learning addresses these limitations by operating on non-Euclidean domains, capturing spatial, topological, and physicochemical features essential to protein function. This perspective outlines the current applications of GDL across stability prediction, functional annotation, molecular interaction modeling, and de novo protein design. We highlight recent methodological advances in model generalization, interpretability, and robustness, particularly under data-scarce conditions. A unified framework is proposed that integrates GDL with explainable AI and structure-based validation to support transparent, autonomous design. As GDL converges with generative modeling and high-throughput experimentation, it is emerging as a central technology in next-generation protein engineering and synthetic biology.

Citations

• Aligning Protein Conformation Ensemble Generation with Physical Feedback
• Efficient GNN Training Through Structure-Aware Randomized Mini-Batching
• Uncertainty Quantification in Graph Neural Networks with Shallow Ensembles
• EquiCPI: SE(3)-Equivariant Geometric Deep Learning for Structure-Aware Prediction of Compound-Protein Interactions
• Learning from spatially inhomogenous data: resolution-adaptive convolutions for multiple sclerosis lesion segmentation
• A Model-Centric Review of Deep Learning for Protein Design
• Scalable emulation of protein equilibrium ensembles with generative deep learning
• From thermodynamics to protein design: Diffusion models for biomolecule generation towards autonomous protein engineering
• Proteins with alternative folds reveal blind spots in AlphaFold-based protein structure prediction
• A Survey of Deep Learning Methods in Protein Bioinformatics and its Impact on Protein Design
• iMFP-LG: Identify Novel Multi-functional Peptides Using Protein Language Models and Graph-based Deep Learning
• Challenge for Deep Learning: Protein Structure Prediction of Ligand-Induced Conformational Changes at Allosteric and Orthosteric Sites
• Large language models and their applications in bioinformatics
• Graph Neural Network Causal Explanation via Neural Causal Models
• Multi-Peptide: Multimodality Leveraged Language-Graph Learning of Peptide Properties
• Graph Neural Network Training Systems: A Performance Comparison of Full-Graph and Mini-Batch
• PILOT: Equivariant diffusion for pocket conditioned de novo ligand generation with multi-objective guidance via importance sampling
• Assessing the potential of deep learning for protein-ligand docking
• Accurate structure prediction of biomolecular interactions with AlphaFold 3
• Enhancing Protein Predictive Models via Proteins Data Augmentation: A Benchmark and New Directions
• Attention Meets Post-hoc Interpretability: A Mathematical Perspective
• PepGB: Facilitating peptide drug discovery via graph neural networks
• ProSTAGE: Predicting Effects of Mutations on Protein Stability by Using Protein Embeddings and Graph Convolutional Networks
• Breaking the Curse of Dimensionality in Deep Neural Networks by Learning Invariant Representations
• Over-Squashing in Graph Neural Networks: A Comprehensive survey
• PeptideBERT: A Language Model based on Transformers for Peptide Property Prediction
• Semantic Interpretation and Validation of Graph Attention-based Explanations for GNN Models
• DeepGATGO: A Hierarchical Pretraining-Based Graph-Attention Model for Automatic Protein Function Prediction
• De novo design of protein structure and function with RFdiffusion
• PIGNet2: A Versatile Deep Learning-based Protein-Ligand Interaction Prediction Model for Binding Affinity Scoring and Virtual Screening
• A Survey on Explainability of Graph Neural Networks
• AbODE: Ab Initio Antibody Design using Conjoined ODEs
• SE(3) diffusion model with application to protein backbone generation
• Ordered GNN: Ordering Message Passing to Deal with Heterophily and Over-smoothing
• Generating Novel, Designable, and Diverse Protein Structures by Equivariantly Diffusing Oriented Residue Clouds
• Evolutionary-scale prediction of atomic-level protein structure with a language model
• Integration of Pre-trained Protein Language Models into Geometric Deep Learning Networks
• Distributed Graph Neural Network Training: A Survey
• Structure-based drug design with geometric deep learning
• Protein Sequence and Structure Co-Design with Equivariant Translation
• Graph neural networks for materials science and chemistry
• Robust deep learning–based protein sequence design using ProteinMPNN
• Tranception: protein fitness prediction with autoregressive transformers and inference-time retrieval
• Protein Representation Learning by Geometric Structure Pretraining
• AlphaDesign: A graph protein design method and benchmark on AlphaFoldDB
• Towards the Explanation of Graph Neural Networks in Digital Pathology with Information Flows
• Iterative Refinement Graph Neural Network for Antibody Sequence-Structure Co-design
• Geometric Deep Learning on Molecular Representations
• Highly accurate protein structure prediction with AlphaFold
• Accurate prediction of protein structures and interactions using a three-track neural network
• Hyperparameter Optimization: Foundations, Algorithms, Best Practices and Open Challenges
• Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gauges
• GraphSVX: Shapley Value Explanations for Graph Neural Networks
• E(n) Equivariant Graph Neural Networks
• On Explainability of Graph Neural Networks via Subgraph Explorations
• CF-GNNExplainer: Counterfactual Explanations for Graph Neural Networks
• Parameterized Explainer for Graph Neural Network
• PGM-Explainer: Probabilistic Graphical Model Explanations for Graph Neural Networks
• GEFA: Early Fusion Approach in Drug-Target Affinity Prediction
• Learning from Protein Structure with Geometric Vector Perceptrons
• Self-Attention Attribution: Interpreting Information Interactions Inside Transformer
• Scattering GCN: Overcoming Oversmoothness in Graph Convolutional Networks
• Ripple Walk Training: A Subgraph-based training framework for Large and Deep Graph Neural Network
• GraphLIME: Local Interpretable Model Explanations for Graph Neural Networks
• ASAP: Adaptive Structure Aware Pooling for Learning Hierarchical Graph Representations
• Evaluating Protein Transfer Learning with TAPE
• Strategies for Pre-training Graph Neural Networks
• ProteinNet: a standardized data set for machine learning of protein structure
• Relational inductive biases, deep learning, and graph networks
• Graph Attention Networks
• Methods for Interpreting and Understanding Deep Neural Networks
• Opening the Black Box of Deep Neural Networks via Information
• Variational Graph Auto-Encoders
• Notes on Kullback-Leibler Divergence and Likelihood
• Biomolecular Simulation: A Computational Microscope for Molecular Biology

Cited by

• Machine Learning-Driven Enzyme Mining: Opportunities, Challenges, and Future Perspectives

Related

• Context-aware geometric deep learning for protein sequence design
• Geometric Deep Learning for Structure-Based Drug Design: A Survey
• Integration of Pre-trained Protein Language Models into Geometric Deep Learning Networks
• Protein engineering in the deep learning era
• Leveraging Deep Generative Model For Computational Protein Design And Optimization