Haoyu Dai
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Haoyu Dai has a total of 85 co-authors. Here are the top co-authors by number of shared publications:
- Brian L. Hie
Arc Institute, Palo Alto, CA, USA
3 shared publications - Luis S. Mille-Fragoso
Department of Bioengineering, Stanford University, Stanford, USA
3 shared publications - Xiaojing J. Gao
Department of Chemical Engineering, Stanford University, Stanford, CA, USA
3 shared publications - Ben Viggiano
Stanford University, Stanford, CA, USA
2 shared publications - Euan Angus Ashley
Stanford University School of Medicine
2 shared publications Jim L. Zhang
Department of Structural Biology, Stanford University, Stanford, CA, USA
2 shared publications
Latest Publications
- 23 June 2026
- Published by Springer Nature in Nature Biotechnology
Efficient generation of epitope-targeted antibodies with Germinal
Obtaining antibodies to specific protein targets is a widely important yet experimentally laborious process. Meanwhile, computational methods for antibody design have been limited by low success rates that require resource-intensive screening. Here we introduce Germinal, a broadly enabling generative pipeline that designs antibodies against specific epitopes with nanomolar binding affinities while requiring only low-n experimental testing. Our method co-optimizes antibody structure and sequence by integrating a structure predictor with an antibody-specific protein language model to perform de novo design of functional complementarity-determining regions onto a user-specified structural framework. When tested against four diverse protein targets, Germinal designed functional antibodies across all targets and binder formats, testing only 43-101 designs for each antigen. Validated designs also exhibited robust expression in mammalian cells and high sequence and structural novelty. We provide open-source code and full computational and experimental protocols to facilitate wide adoption.
- 23 June 2026
- Published by Springer Nature in Nature Biotechnology
Efficient generation of epitope-targeted antibodies with Germinal
Obtaining antibodies to specific protein targets is a widely important yet experimentally laborious process. Meanwhile, computational methods for antibody design have been limited by low success rates that require resource-intensive screening. Here we introduce Germinal, a broadly enabling generative pipeline that designs antibodies against specific epitopes with nanomolar binding affinities while requiring only low-n experimental testing. Our method co-optimizes antibody structure and sequence by integrating a structure predictor with an antibody-specific protein language model to perform de novo design of functional complementarity-determining regions onto a user-specified structural framework. When tested against four diverse protein targets, Germinal designed functional antibodies across all targets and binder formats, testing only 43-101 designs for each antigen. Validated designs also exhibited robust expression in mammalian cells and high sequence and structural novelty. We provide open-source code and full computational and experimental protocols to facilitate wide adoption.
- Top Cited PapersOpen Access
- 4 March 2026
- Published by Springer Nature in Nature
- Vol. 652 (8112) , 1-13
Genome modelling and design across all domains of life with Evo 2
All of life encodes information with DNA. Although tools for genome sequencing, synthesis and editing have transformed biological research, we still lack sufficient understanding of the immense complexity encoded by genomes to predict the effects of many classes of genomic changes or to intelligently compose new biological systems. Artificial intelligence models that learn information from genomic sequences across diverse organisms have increasingly advanced prediction and design capabilities1,2. Here we introduce Evo 2, a biological foundation model trained on 9 trillion DNA base pairs from a highly curated genomic atlas spanning all domains of life to have a 1 million token context window with single-nucleotide resolution. Evo 2 learns to accurately predict the functional impacts of genetic variation—from noncoding pathogenic mutations to clinically significant BRCA1 variants—without task-specific fine-tuning. Mechanistic interpretability analyses reveal that Evo 2 learns representations associated with biological features, including exon–intron boundaries, transcription factor binding sites, protein structural elements and prophage genomic regions. The generative abilities of Evo 2 produce mitochondrial, prokaryotic and eukaryotic sequences at genome scale with greater naturalness and coherence than previous methods. Evo 2 also generates experimentally validated chromatin accessibility patterns when guided by predictive models3,4 and inference-time search. We have made Evo 2 fully open, including model parameters, training code5, inference code and the OpenGenome2 dataset, to accelerate the exploration and design of biological complexity.
- 4 March 2026
- Published by Springer Nature in Nature
- Vol. 652 (8112) , 1-13
Top Cited PapersOpen AccessGenome modelling and design across all domains of life with Evo 2
All of life encodes information with DNA. Although tools for genome sequencing, synthesis and editing have transformed biological research, we still lack sufficient understanding of the immense complexity encoded by genomes to predict the effects of many classes of genomic changes or to intelligently compose new biological systems. Artificial intelligence models that learn information from genomic sequences across diverse organisms have increasingly advanced prediction and design capabilities1,2. Here we introduce Evo 2, a biological foundation model trained on 9 trillion DNA base pairs from a highly curated genomic atlas spanning all domains of life to have a 1 million token context window with single-nucleotide resolution. Evo 2 learns to accurately predict the functional impacts of genetic variation—from noncoding pathogenic mutations to clinically significant BRCA1 variants—without task-specific fine-tuning. Mechanistic interpretability analyses reveal that Evo 2 learns representations associated with biological features, including exon–intron boundaries, transcription factor binding sites, protein structural elements and prophage genomic regions. The generative abilities of Evo 2 produce mitochondrial, prokaryotic and eukaryotic sequences at genome scale with greater naturalness and coherence than previous methods. Evo 2 also generates experimentally validated chromatin accessibility patterns when guided by predictive models3,4 and inference-time search. We have made Evo 2 fully open, including model parameters, training code5, inference code and the OpenGenome2 dataset, to accelerate the exploration and design of biological complexity.
- Open Access
Efficient generation of epitope-targeted de novo antibodies with Germinal
Obtaining novel antibodies against specific protein targets is a widely important yet experimentally laborious process. Meanwhile, computational methods for antibody design have been limited by low success rates that currently require resource-intensive screening. Here, we introduce Germinal, a broadly enabling generative pipeline that designs antibodies against specific epitopes with nanomolar binding affinities while requiring only low-n experimental testing. Our method co-optimizes antibody structure and sequence by integrating a structure predictor with an antibody-specific protein language model to perform de novo design of functional complementarity-determining regions (CDRs) onto a user-specified structural framework. When tested against four diverse protein targets, Germinal successfully designed functional antibodies across all targets and binder formats, testing only 43-101 designs for each antigen. Validated designs also exhibited robust expression in mammalian cells and high sequence and structural novelty. We provide open-source code and full computational and experimental protocols to facilitate wide adoption. Germinal represents a milestone in efficient, epitope-targeted de novo antibody design, with notable implications for the development of molecular tools and therapeutics.
Open AccessEfficient generation of epitope-targeted de novo antibodies with Germinal
Obtaining novel antibodies against specific protein targets is a widely important yet experimentally laborious process. Meanwhile, computational methods for antibody design have been limited by low success rates that currently require resource-intensive screening. Here, we introduce Germinal, a broadly enabling generative pipeline that designs antibodies against specific epitopes with nanomolar binding affinities while requiring only low-n experimental testing. Our method co-optimizes antibody structure and sequence by integrating a structure predictor with an antibody-specific protein language model to perform de novo design of functional complementarity-determining regions (CDRs) onto a user-specified structural framework. When tested against four diverse protein targets, Germinal successfully designed functional antibodies across all targets and binder formats, testing only 43-101 designs for each antigen. Validated designs also exhibited robust expression in mammalian cells and high sequence and structural novelty. We provide open-source code and full computational and experimental protocols to facilitate wide adoption. Germinal represents a milestone in efficient, epitope-targeted de novo antibody design, with notable implications for the development of molecular tools and therapeutics.
- Open Access
Generative steering to adapt proteins beyond native biological contexts
Proteins evolved for native biological contexts often fail when deployed in new cellular environments or therapeutic settings. We present ProVADA, a test-time framework that steers a pretrained protein generator toward externally defined properties while preserving functional constraints. Its Mixture-Adaptation Directed Annealing sampler combines coordinated mutations, population selection, and annealing to navigate rugged objective landscapes. Experimentally, ProVADA converted secreted human renin into an active, drug-inhibitable cytosolic protease and extensively deimmunized two murine single-chain antibody fragments to reduce HLA visibility while retaining antigen recognition. ESCAPE-seq measurements confirmed substantial depletion of HLA-I binder peptides across alleles. Deimmunized FMC63 variants retained CAR-T-cell killing and showed fewer binders than a tested humanized comparator. Together, these results demonstrate that broad sequence redesign can adapt functional proteins to non-native biological contexts.
Open AccessGenerative steering to adapt proteins beyond native biological contexts
Proteins evolved for native biological contexts often fail when deployed in new cellular environments or therapeutic settings. We present ProVADA, a test-time framework that steers a pretrained protein generator toward externally defined properties while preserving functional constraints. Its Mixture-Adaptation Directed Annealing sampler combines coordinated mutations, population selection, and annealing to navigate rugged objective landscapes. Experimentally, ProVADA converted secreted human renin into an active, drug-inhibitable cytosolic protease and extensively deimmunized two murine single-chain antibody fragments to reduce HLA visibility while retaining antigen recognition. ESCAPE-seq measurements confirmed substantial depletion of HLA-I binder peptides across alleles. Deimmunized FMC63 variants retained CAR-T-cell killing and showed fewer binders than a tested humanized comparator. Together, these results demonstrate that broad sequence redesign can adapt functional proteins to non-native biological contexts.