Abstract:The development of advanced deep learning methods has revolutionized computational protein design. Although the success rate of design has been significantly increased, the overall accuracy of design remains low. Many computational sequence design approaches are devoted to recover the original sequences for given protein structures by encoding the environment of the central residue without considering atomic details of side chains. This may limit the exploration of new sequences that can fold into the same structure and restrain function design that depends on interaction details. In this study, we proposed a novel deep learning frame-work, GeoSeqBuilder, to learn the relationship between protein structure and sequence based on rotational and translational invariance by extracting the information from relative locations. We utilized geometric deep learning to fetch the spatial local geometric features from protein backbones and explicitly incorporated three-body interactions to learn the inter-residue coupling information, and then determined the central residue type. Our model recovers over 50% native residue types and simultaneously gives highly accurate prediction of side-chain conformations which gives the atomic interaction details and circumvents the dependence of protein structure prediction tools. We used the likelihood confidence log as scoring function for sequence and structure consistence evaluation which exhibits strong correlation with TM-score, and can be applied to recognize near-native structures from protein decoys pool in protein structure prediction. We have used GeoSeqBuilder to design sequences for two proteins, including thiore-doxin and a hallucinated protein. All of the 15 sequences experimentally tested can be expressed as soluble monomeric proteins with high thermal stability and correct secondary structures. We further solved one crystal structure for thioredoxin and two for the hallucinated structure and all the experimentally solved structures are in good agreement with the designed models. The two designed sequences for the hallucination structure are novel without any homologous sequences within the latest released database clust30. The ability of GeoSeqBuilder to design new sequences for given protein structures with atomic details makes it applicable, not only for sequence design, but also for protein-protein interaction and functional protein design.

Scalable protein design using optimization in a relaxed sequence space

Global-Context Aware Generative Protein Design

Generative De Novo Protein Design with Global Context

Protein Design by Integrating Machine Learning with Quantum Annealing and Quantum-inspired Optimization

Protein sequence design with a learned potential

Robust deep learning based protein sequence design using ProteinMPNN

All-atom protein sequence design based on geometric deep learning

Protein sequence design on given backbones with deep learning

Highly accurate and robust protein sequence design with CarbonDesign

Protein Language Model Supervised Precise and Efficient Protein Backbone Design Method

From sequence to function through structure: deep learning for protein design

Accurate and efficient protein sequence design through learning concise local environment of residues

Protein sequence design by explicit energy landscape optimization

OPUS-Design: Designing Protein Sequence from Backbone Structure with 3DCNN and Protein Language Model

An integrative approach to protein sequence design through multiobjective optimization

PRO-LDM: Protein Sequence Generation with a Conditional Latent Diffusion Model

Scalable design of repeat protein structural dynamics via probabilistic coarse-grained models

Generating Novel, Designable, and Diverse Protein Structures by Equivariantly Diffusing Oriented Residue Clouds

A new age in protein design empowered by deep learning

SPDesign: protein sequence designer based on structural sequence profile using ultrafast shape recognition

ZetaDesign: an end-to-end deep learning method for protein sequence design and side-chain packing