A Generative Neuro‐Symbolic AI for Protein Sequence Design

Abstract

Deep learning has revolutionized computational protein design, enabling the generation of sequences that fold onto target backbones with unprecedented accuracy. However, state-of-the-art inverse folding tools largely rely on auto-regressive sampling. While powerful, this paradigm is increasingly recognized for its inability to "think ahead", a crucial capacity to reliably create the complex, long-range inter-residue dependencies essential for most biological functions. To overcome these fundamental limitations, we introduced EffieDes , a generative neuro-symbolic AI framework that synergizes the predictive capabilities of deep learning with the logical precision of automated reasoning. EffieDes leverages deep learning to encode the target backbone's fitness landscape into Effie -a fully decomposable probabilistic graphical model (Potts model). This landscape can then be rigorously explored by an automated reasoning prover to identify sequences that simultaneously satisfy complex design constraints and optimize backbone fitness. We validated this neuro-symbolic approach through the design of orthogonal sequence pairs that adopt identical folds but exhibit selective self-assembly, as well as the design of a de novo selective nanobody with nanomolar affinity for an immune-evasive SARS-CoV-2 variant. EffieDes provides a robust architecture for precisely dissecting learned fitness landscapes, offering a new path toward proteins with highly optimized performances and sophisticated functional objectives.

Generative AI for controllable protein sequence design: A survey
Yiheng Zhu, Zitai Kong, Jialu Wu +7
npj Drug Discovery.3 Jul 2026Peer-reviewedCC BY-NC-NDFull text8 citedMolecular biology & gene regulationOpenAlex
EquiDesign: a versatile computational framework for de novo protein sequence design.
Qiang Gao, Zhongyu Xu, Kai Chen +4
Briefings in bioinformatics1 Sep 2026Peer-reviewedCC BYFull textMolecular biology & gene regulationEurope PMC
Joint protein, mRNA and DNA sequence design and optimization with nucleotide-level Potts models
B. Banaszewski, L. J. Dornfeld, D. Chen +2
bioRxiv5 Oct 2026PreprintCC BYFull textBiophysicsbioRxiv
Comparative Conformational Sampling of Apo Adenylate Kinase by Enhanced Molecular Dynamics and a Sequence-Based Generative Ensemble Model
Qingling Zhao, Ting Cui, Ying Li
Molecules1 Oct 2026Peer-reviewedCC BY 4.0Full textMolecular biology & gene regulationCrossrefEurope PMC
Representation learning for low-data protein fitness prediction using sequence–structure embeddings and graph neural networks
Chanoknun Phunnasorn, Narupat Hongdilokkul, Wasakorn Laesanklang +2
Scientific Reports16 Oct 2026Peer-reviewedCC BY-NC-ND 4.0Full textMolecular biology & gene regulationCrossref

Cite

BibTeX RIS

BibTeX

@article{defresne2026generative,
  title = {{A Generative Neuro‐Symbolic AI for Protein Sequence Design}},
  author = {Marianne Defresne and Delphine Dessaux and Samuel Buchet and Lucie Barthe and Liza Ammar‐khodja and Bessam Azizi and Valentin Durante and Gianluca Cioci and Simon de Givry and Alain Roussel and Luis F Garcia‐alles and Thomas Schiex and Sophie Barbe},
  journal = {Advanced Science},
  year = {2026},
  month = jul,
  doi = {10.1002/advs.76464},
  url = {https://hal.science/hal-05776660v1},
  note = {CC BY 4.0},
}

RIS

TY  - JOUR
TI  - A Generative Neuro‐Symbolic AI for Protein Sequence Design
AU  - Marianne Defresne
AU  - Delphine Dessaux
AU  - Samuel Buchet
AU  - Lucie Barthe
AU  - Liza Ammar‐khodja
AU  - Bessam Azizi
AU  - Valentin Durante
AU  - Gianluca Cioci
AU  - Simon de Givry
AU  - Alain Roussel
AU  - Luis F Garcia‐alles
AU  - Thomas Schiex
AU  - Sophie Barbe
PY  - 2026
DA  - 2026/07/30
JO  - Advanced Science
DO  - 10.1002/advs.76464
UR  - https://hal.science/hal-05776660v1
AB  - Deep learning has revolutionized computational protein design, enabling the generation of sequences that fold onto target backbones with unprecedented accuracy. However, state-of-the-art inverse folding tools largely rely on auto-regressive sampling. While powerful, this paradigm is increasingly recognized for its inability to "think ahead", a crucial capacity to reliably create the complex, long-range inter-residue dependencies essential for most biological functions. To overcome these fundamental limitations, we introduced EffieDes , a generative neuro-symbolic AI framework that synergizes the predictive capabilities of deep learning with the logical precision of automated reasoning. EffieDes leverages deep learning to encode the target backbone's fitness landscape into Effie -a fully decomposable probabilistic graphical model (Potts model). This landscape can then be rigorously explored by an automated reasoning prover to identify sequences that simultaneously satisfy complex design constraints and optimize backbone fitness. We validated this neuro-symbolic approach through the design of orthogonal sequence pairs that adopt identical folds but exhibit selective self-assembly, as well as the design of a de novo selective nanobody with nanomolar affinity for an immune-evasive SARS-CoV-2 variant. EffieDes provides a robust architecture for precisely dissecting learned fitness landscapes, offering a new path toward proteins with highly optimized performances and sophisticated functional objectives.
SN  - 2198-3844
ER  -

Abstract

Related

Cite