Biology subjects

Brixi, G.

Publications and source records attributed to Brixi, G..

12 recordsLinked to original sources

bioRxiv2026

Few-shot regulatory sequence design with a genomic language model

A central challenge in generative biology is translating knowledge learned from natural sequences and experiments into designs with desired functions. In-context learning (ICL) is an emergent capability of large language models that allows them to adapt to new tasks from a few examples without additional training, enabling these models to become general-purpose learners. Here, we show that the genome language model, Evo 2, exhibits generative ICL in biological sequence design. Prompting with just tens of examples conditions Evo 2 to generate novel and diverse regulatory DNA sequences without any additional training. Mechanistic analysis suggests that genomic ICL may originate from induction, similar to phenomena observed in natural language. Reinforcement learning leveraging sequence-to-function models as experiment-derived rewards further strengthens ICL. We first validate this approach by generating enhancers that drive cell-type-specific gene expression in three human cell lines. Extending this approach in vivo, enhancers designed using human regulatory element data selectively label layer 5 extratelencephalic neurons in the mouse cortex with high specificity and stronger activity than existing enhancers. Our study provides evidence of emergent properties in genomic sequence models and introduces a promising approach for improving functional conditioning in generative biology.

bioengineering↗

bioRxiv2026

Efficient biological design via genomic in-context learning

In-context learning (ICL), the ability to perform new tasks from a few examples supplied in a prompt, is a defining emergent capability of large language models. Here, we show that ICL emerges in the genome language model Evo 2 and enables biological design. With as few as four examples in the prompt and no additional training, Evo 2 generates realistic metal-binding proteins, de novo antibodies, protein-RNA systems, and entire bacteriophage genomes. For {Phi}X174-like bacteriophage design, ICL surpasses previously fine-tuned models on computational measures of design quality and diversity. Beyond {Phi}X174, conditioning the model with different phage prompts enables us to efficiently design MS2 and M13 genomes in silico without fine-tuning. Iteratively updating the prompt with newly generated phage genomes, which we refer to as genomic chain-of-thought, further improves the diversity of designs while maintaining quality metrics. Paired with an inference engine that accelerates generation up to 35-fold and extends practicable context to a megabase, these results establish ICL as a versatile and efficient paradigm for biological sequence design.

bioinformatics↗

bioRxiv2026

Coevolutionary mining of prokaryotic non-coding elements with a genome language model

Microbial genomes encode compact molecular machines and diverse non-coding RNAs (ncRNAs) essential to gene regulation, pathogenesis, and many foundational biotechnologies. However, annotation remains largely protein-centric and homology-driven. Here, we introduce Minerva, a framework for coevolutionary mining that uses genome language models to predict local interactions directly from sequence as two-dimensional maps. Introducing two complementary techniques, categorical Jacobian fingerprinting and interaction heads, we demonstrate fast, accurate, alignment-free prediction of ncRNA base-pairing, monomeric protein contacts, and repetitive sequence motifs. Applied to 150 bacterial genomes, Minerva recovers known systems and predicts that 84.3% of predicted intergenic base-pairing falls outside of known annotations. In Pseudomonas, we find that the widespread TwoAYGGAY ncRNA family carries large secondary-structure extensions and is often flanked by short upstream repetitive motifs and larger downstream genomic repeats. Interpreting the coevolution maps, we find that Minerva emergently detects open-reading-frame (ORF) signatures at the DNA level despite never being trained to do so. In prophages within these genomes, we discover that Unknown Group 27 (UG27) reverse transcriptase systems encode arrays of structurally conserved yet sequence-diverse ncRNAs that template complementary DNA (cDNA) hairpin products. Together, these results establish coevolutionary mining as a scalable route to genome annotation and biological discovery across the rapidly expanding microbial universe.

bioinformatics↗

bioRxiv2026

Forecasting viral evolution from phylogenetic trees

Viral mutation forecasting plays a key role in pandemic preparedness by enabling researchers to anticipate novel variants and design proactive interventions. Evolutionary histories, represented as phylogenetic trees, offer key insights into the emergence of past and present strains, yet their role in predicting future sequence changes remains largely unexplored. We introduce antiGen, a machine learning model that forecasts the evolutionary future of viruses by learning from their evolutionary past. antiGen achieves state-of-the-art performance for predicting mutations to the SARS-CoV-2 spike protein, anticipating never-before-seen mutations and mutations that emerge years after the model's training window. antiGen-forecasted spike mutations also retain pseudoviral infectivity in vitro. Moreover, antiGen demonstrates leading predictive performance on surface proteins of influenza virus, respiratory syncytial virus, and dengue virus despite far less available sequencing data. Viral evolution models that explicitly learn from phylogenetic structure offer a valuable resource for applications ranging from epidemiological modeling to therapeutic development.

evolutionary biology↗

bioRxiv2025

Generative design of novel bacteriophages with genome language models

Many important biological functions arise not from single genes, but from complex interactions encoded by entire genomes. Genome language models have emerged as a promising strategy for designing biological systems, but their ability to generate functional sequences at the scale of whole genomes has remained untested. Here, we report the first generative design of viable bacteriophage genomes. We leveraged frontier genome language models, Evo 1 and Evo 2, to generate whole-genome sequences with realistic genetic architectures and desirable host tropism, using the lytic phage {Phi}X174 as our design template. Experimental testing of AI-generated genomes yielded 16 viable phages with substantial evolutionary novelty. Cryo-electron microscopy revealed that one of the generated phages utilizes an evolutionarily distant DNA packaging protein within its capsid. Multiple phages demonstrate higher fitness than {Phi}X174 in growth competitions and in their lysis kinetics. A cocktail of the generated phages rapidly overcomes {Phi}X174-resistance in three E. coli strains, demonstrating the potential utility of our approach for designing phage therapies against rapidly evolving bacterial pathogens. This work provides a blueprint for the design of diverse synthetic bacteriophages and, more broadly, lays a foundation for the generative design of useful living systems at the genome scale.

synthetic biology↗

bioRxiv2025

The Dayhoff Atlas: scaling sequence diversity for improved protein generation

Organized information powers modern biology, a framework pioneered by Margaret Dayhoffs Atlas of Protein Sequence and Structure and advanced by todays databases and computational methods. Here, we extend this paradigm for the AI era, presenting the Dayhoff Atlas of protein sequence data and generative models to accelerate protein biology and design. The Atlas introduces GigaRef, the largest open dataset of natural proteins, spanning 3.34B genomic and metagenomic sequences across 1.70B clusters, and BackboneRef, which distills structural information from 240,811 synthetic backbones into 46M synthetic sequences. Leveraging these datasets, we trained the Dayhoff protein language models, which can predict mutation effects, scaffold structural motifs, and generate novel proteins within families. Training on metagenomic and structure-based synthetic sequences increased the expression rates of generated proteins, demonstrating the value of data diversity and scale. We release the Dayhoff Atlas code, datasets, and models under a permissive license to empower computation in protein design.

bioengineering↗

bioRxiv2025

Genome modeling and design across all domains of life with Evo 2

All of life encodes information with DNA. While tools for sequencing, synthesis, and editing of genomic code have transformed biological research, intelligently composing new biological systems would also require a deep understanding of the immense complexity encoded by genomes. We introduce Evo 2, a biological foundation model trained on 9.3 trillion DNA base pairs from a highly curated genomic atlas spanning all domains of life. We train Evo 2 with 7B and 40B parameters to have an unprecedented 1 million token context window with single-nucleotide resolution. Evo 2 learns from DNA sequence alone to accurately predict the functional impacts of genetic variation--from noncoding pathogenic mutations to clinically significant BRCA1 variants--without task-specific finetuning. Applying mechanistic interpretability analyses, we reveal that Evo 2 autonomously learns a breadth of biological features, including exon-intron boundaries, transcription factor binding sites, protein structural elements, and prophage genomic regions. Beyond its predictive capabilities, Evo 2 generates mitochondrial, prokaryotic, and eukaryotic sequences at genome scale with greater naturalness and coherence than previous methods. Guiding Evo 2 via inference-time search enables controllable generation of epigenomic structure, for which we demonstrate the first inference-time scaling results in biology. We make Evo 2 fully open, including model parameters, training code, inference code, and the OpenGenome2 dataset, to accelerate the exploration and design of biological complexity.

genomics↗

bioRxiv2024

Protein language models learn evolutionary statistics of interacting sequence motifs

Protein language models (pLMs) have emerged as potent tools for predicting and designing protein structure and function, and the degree to which these models fundamentally understand the inherent biophysics of protein structure stands as an open question. Motivated by a discovery that pLM-based structure predictors erroneously predict nonphysical structures for protein isoforms, we investigated the nature of sequence context needed for contact predictions in the pLM ESM-2. We demonstrate by use of a "categorical Jacobian" calculation that ESM-2 stores statistics of coevolving residues, analogously to simpler modelling approaches like Markov Random Fields and Multivariate Gaussian models. We further investigated how ESM-2 "stores" information needed to predict contacts by comparing sequence masking strategies, and found that providing local windows of sequence information allowed ESM-2 to best recover predicted contacts. This suggests that pLMs predict contacts by storing motifs of pairwise contacts. Our investigation highlights the limitations of current pLMs and underscores the importance of understanding the underlying mechanisms of these models. Significance StatementProtein language models (pLMs) have exhibited remarkable capabilities in protein structure prediction and design. However, the extent to which they comprehend the intrinsic biophysics of protein structures remains uncertain. We present a suite of analyses that dissect how the flagship pLM ESM-2 predicts structure. Motivated by a consistent error of protein isoforms predicted as structured fragments, we developed a completely unsupervised method to uniformly evaluate any protein language model that allows for us to compare coevolutionary statistics to older linear models. We further identified t hat E SM-2 a ppears to have a precise context size that is needed to predict inter-residue contacts. Our study highlights the current limitations of pLMs and contributes to a deeper understanding of their underlying mechanisms, paving the way for more reliable protein structure predictions.

bioinformatics↗

bioRxiv2023

Transferrin Receptor Targeting Chimeras (TransTACs) for Membrane Protein Degradation

Cancer cells require high levels of iron for rapid proliferation, leading to a significant upregulation of the iron carrier protein Transferrin Receptor (TfR) on their cell surface. Leveraging this phenomenon and the exceptionally fast endocytosis rate of TfR, we introduce Transferrin Receptor TArgeting Chimeras (TransTAC), a novel molecular archetype for membrane protein degradation in cancers and other cell types. TransTACs repurpose the naturally recycling receptor TfR1 for protein degradation. To accomplish this, we utilized a combination of protein engineering strategies to redirect the target protein from recycling-endosome trafficking to lysosomal degradation. We show that TransTACs can highly efficiently degrade a diverse range of single-pass, multi-pass, native, or synthetic membrane proteins, establishing new possibilities for targeted cancer therapy.

bioengineering↗

bioRxiv2022

Efficient Human Germ Cell Specification from Stem Cells via Combinatorial Expression of Transcription Factors

The generation of germline cell types from human induced pluripotent stem cells (hiPSCs) represents a key milestone toward in vitro gametogenesis, which has the potential to transform reproductive modeling and medicine. Methods to recapitulate advanced germline cell specification in vitro have relied on extensive, long term culture methods, the most notable of which is a four-month culture protocol employing xenogeneic reconstituted ovaries with mouse embryonic ovarian somatic cells. Recently, transcription factor (TF)-based methods have demonstrated the feasibility of exogenous factor expression to directly differentiate hiPSCs into cell types of interest, including various ovarian cell types. The protocols leveraged in these studies, however, utilize more local methods of factor selection, such as basic differential gene expression analysis, and lower-throughput screening strategies via iterative testing of a small set of TFs. In this work, we integrate our recently-described graph theory pipeline and highly-parallelized screening protocols to globally identify and screen 46 oogenesis-regulating TFs for their role in human germline formation. We identify ZNF281, LHX8, SOHLH1, ZGLP1, and ANHX whose combinatorial overexpression drives DDX4+ induced oogonia-like cell (iOLC) formation from hiPSCs. In contrast to previous methods, our protocol employs a simple four-day, feeder-free monolayer culture condition. We additionally demonstrate a method of post-isolation, feeder-free expansion of DDX4+ iOLCs that shows retained cell identity in vitro. We additionally identify DLX5, HHEX, and FIGLA whose individual overexpression enhances hPGCLC formation from hiPSCs. We characterize these TF-based iOLCs and hPGCLCs via gene and protein expression analyses and demonstrate their broad similarity to in vivo and in vitro-derived oogonia and primordial germ cells. Together, these results identify new regulatory factors that enhance in vitro human germ cell specification and further establish unique computational and experimental tools for human in vitro oogenesis research.

developmental biology↗

bioRxiv2022

Directed Differentiation of Human iPSCs to Functional Ovarian Granulosa-Like Cells via Transcription Factor Overexpression

An in vitro model of human ovarian follicles would greatly benefit the study of female reproduction. Ovarian development requires the combination of germ cells and their supporting somatic cells, known as granulosa cells. Whereas efficient protocols exist for generating human primordial germ cell-like cells (hPGCLCs) from human iPSCs, a method of generating granulosa cells has been elusive. Here we report that simultaneous overexpression of two transcription factors (TFs) can direct the differentiation of human iPSCs to granulosa-like cells. We elucidate the regulatory effects of several granulosa-related TFs, and establish that overexpression of NR5A1 and either RUNX1 or RUNX2 is necessary and sufficient to generate granulosa-like cells. Our granulosa-like cells form ovary-like organoids (ovaroids) when aggregated with hPGCLCs, and recapitulate key ovarian phenotypes including support of germ cell maturation, follicle formation, and steroidogenesis. This model system will provide unique opportunities for studying human ovarian biology, and may enable the development of therapies for female reproductive health.

developmental biology↗

bioRxiv2022

Design of Peptide-Based Protein Degraders via Contrastive Deep Learning

AO_SCPLOWBSTRACTC_SCPLOWTherapeutic modalities targeting pathogenic proteins are the gold standard of treatment for multiple disease indications. Unfortunately, a significant portion of these proteins are considered "undruggable" by standard small molecule-based approaches, largely due to their disordered nature and instability. Designing functional peptides to undruggable targets, either as standalone binders or fusions to effector domains, thus presents a unique opportunity for therapeutic intervention. In this work, we adapt recent models for contrastive language-image pre-training (CLIP) to devise a unified, sequence-based framework to design target-specific peptides. Furthermore, by leveraging known experimental binding proteins as scaffolds, we create a streamlined inference pipeline, termed Cut&CLIP, that efficiently selects peptides for downstream screening. Finally, we experimentally fuse candidate peptides to E3 ubiquitin ligase domains and demonstrate robust intracellular degradation of pathogenic protein targets in human cells, motivating further development of our technology for future clinical translation.

synthetic biology↗