Pre-training Genomic Language Model with Variants for Better Modeling Functional Genomics.

Sequence-to-function models can predict gene expression from sequence data and be used to link genetic information with transcriptomics data to understand regulatory processes and their effects on complex phenotypes. The genomic language models are pre-trained with large-scale DNA sequences and can generate robust representations of these sequences by learning the genomic context. However, few studies can estimate the predictability of gene expression levels and bridge these two classes of models together to explore individualized gene expression prediction.

GAME: Genomic API for Model Evaluation.

The rapid expansion of genomics datasets and the application of machine learning has produced sequence-to-activity genomics models with ever-expanding capabilities. However, benchmarking these models on practical applications has been challenging because individual projects evaluate their models in ad hoc ways, and there is substantial heterogeneity of both model architectures and benchmarking tasks. To address this challenge, we have created GAME, a system for large-scale, community-led standardized model benchmarking on user-defined evaluation tasks.

DNA assembly for targeted genomic integration and rearrangement in human cells.

Although therapeutic genome editing holds great potential to remedy diverse inherited and acquired disorders, targeted installation of medium to large sized genomic modifications in therapeutically relevant cells remains challenging. We have developed an approach that permits DNA sequence assembly and integration in human cells leveraging CRISPR-targeted dual flap synthesis. This method, named prime assembly, allows for RNA-programmable site-specific integration of single- or double-stranded DNA fragments.

A Molecular Glue Approach to Control the Half-Life of CRISPR-Based Technologies.

Cas9 is a programmable nuclease that has furnished transformative technologies, including base editors and transcription modulators (e.g., CRISPRi/a), but several applications of these technologies, including therapeutics, mandatorily require precision control of their half-life.

Spatial multiomics decipher fibroblast-macrophage dynamics in systemic sclerosis.

Objectives: Stromal-immune crosstalk shapes the pathogenic microenvironment of systemic sclerosis (SSc), but the spatial regulatory networks underlying fibrogenesis remain poorly defined. We aimed to explore tissue organisation and cell coordination in SSc skin, providing spatiotemporal insights into disease mechanisms and bridging the gap between omics discovery and precision medicine.

Methods: We performed spatial transcriptomics on skin biopsies from 10 patients with diffuse cutaneous SSc and 4 healthy controls using the 10× Visium platform.

Custom CRISPR-Cas9 PAM variants via scalable engineering and machine learning.

Engineering and characterizing proteins can be time-consuming and cumbersome, motivating the development of generalist CRISPR-Cas enzymes to enable diverse genome-editing applications. However, such enzymes have caveats such as an increased risk of off-target editing. Here, to enable scalable reprogramming of Cas9 enzymes, we combined high-throughput protein engineering with machine learning to derive bespoke editors that are more uniquely suited to specific targets.

Deaminase-mediated chromatin accessibility profiling with single-allele resolution.

Binding of transcription factors (TFs) at gene regulatory elements controls cellular epigenetic state and gene expression. Current genome-wide chromatin profiling approaches have inherently limited resolution, complicating assessment of TF occupancy and co-occupancy, especially at individual alleles. In this work, we introduce Accessible Chromatin by Cytosine Editing Site Sequencing with ATAC-seq (ACCESS-ATAC), which harnesses a double-stranded DNA cytosine deaminase (Ddd) enzyme to stencil TF binding locations within accessible chromatin regions.

Gene editing without ex vivo culture evades genotoxicity in human hematopoietic stem cells.

Gene editing the BCL11A erythroid enhancer is a validated approach to fetal hemoglobin (HbF) induction for β-hemoglobinopathy therapy, though heterogeneity in edit allele distribution and HbF response may impact its safety and efficacy. Here, we compare combined CRISPR-Cas9 editing of the BCL11A +58 and +55 enhancers with leading gene modification approaches under clinical investigation. Dual targeting of the BCL11A +58 and +55 enhancers with 3xNLS-SpCas9 and two single guide RNAs (sgRNAs) resulted in superior HbF induction, including in sickle cell disease (SCD) patient xenografts, attributable to simultaneous disruption of core half E-box/GATA motifs at both enhancers.

Statistical inference with a manifold-constrained RNA velocity model uncovers cell cycle speed modulations.

Across biological systems, cells undergo coordinated changes in gene expression, resulting in transcriptome dynamics that unfold within a low-dimensional manifold. While low-dimensional dynamics can be extracted using RNA velocity, these algorithms can be fragile and rely on heuristics lacking statistical control. Moreover, the estimated vector field is not dynamically consistent with the traversed gene expression manifold.

CRISPR-CLEAR: Nucleotide-Resolution Mapping of Regulatory Elements via Allelic Readout of Tiled Base Editing.

CRISPR tiling screens have advanced the identification and characterization of regulatory sequences but are limited by low resolution arising from the indirect readout of editing via guide RNA sequencing. This study introduces , an end-to-end experimental assay and computational pipeline, which leverages targeted sequencing of CRISPR-introduced alleles at the endogenous target locus following dense base-editing mutagenesis. This approach enables the dissection of regulatory elements at nucleotide resolution, facilitating a direct assessment of genotype-phenotype effects.

Selective Enhancer Gain-of-Function Deregulates MYC Expression in Multiple Myeloma.

MYC deregulation occurs in the majority of multiple myeloma cases and is associated with progression and worse prognosis. Enhanced MYC expression occurs in about 70% of patients with multiple myeloma, but it is known to be driven by translocation or amplification events in only ∼40% of myelomas. Here, we used CRISPR interference to uncover an epigenetic mechanism of MYC regulation whereby increased accessibility of a plasma cell-type-specific enhancer leads to increased MYC expression.

Scalable assessment of genome editing off-targets associated with genetic variants.

Genome editing with RNA-guided DNA binding factors carries risk of off-target editing at homologous sequences. Genetic variants may introduce sequence changes that increase homology to a genome editing target, thereby increasing risk of off-target editing. Conventional methods to verify candidate off-targets rely on access to cells with genomic DNA carrying these sequences.

EPInformer: a scalable deep learning framework for gene expression prediction by integrating promoter-enhancer sequences with multimodal epigenomic data.

Transcriptional regulation, critical for cellular differentiation and adaptation to environmental changes, involves coordinated interactions among DNA sequences, regulatory proteins, and chromatin architecture. Despite extensive data from consortia like ENCODE, understanding the dynamics of cis-regulatory elements (CREs) in gene expression remains challenging. Deep learning is a powerful tool for learning gene expression and epigenomic signals from DNA sequences, exhibiting superior performance compared to conventional machine learning approaches.

CRISPR PERSIST-On enables heritable and fine-tunable human gene activation.

Current technologies for upregulation of endogenous genes use targeted artificial transcriptional activators but stable gene activation requires persistent expression of these synthetic factors. Although general "hit-and-run" strategies exist for inducing long-term silencing of endogenous genes using targeted artificial transcriptional repressors, to our knowledge no equivalent approach for gene activation has been described to date. Here we show stable gene activation can be achieved by harnessing endogenous transcription factors ( s) that are normally expressed in human cells.

MOSAIC enables saturation mutagenesis of genes and CRISPR prime editing guide RNA optimization in human cells.

CRISPR prime editing offers unprecedented versatility and precision for the installation of genetic edits . Here we describe the development and characterization of the Multiplexing Of Site-specific Alterations for Characterization ( ) method, which leverages a non-viral PCR-based prime editing method to enable rapid installation of thousands of defined edits in pooled fashion. We show that MOSAIC can be applied to perform saturation mutagenesis screens of: (1) the fusion gene, successfully identifying known and potentially new imatinib drug-resistance variants; and (2) the untranslated region (UTR), re-confirming non-coding regulatory elements involved in transcriptional initiation.

Joint genotypic and phenotypic outcome modeling improves base editing variant effect quantification.

CRISPR base editing screens enable analysis of disease-associated variants at scale; however, variable efficiency and precision confounds the assessment of variant-induced phenotypes. Here, we provide an integrated experimental and computational pipeline that improves estimation of variant effects in base editing screens. We use a reporter construct to measure guide RNA (gRNA) editing outcomes alongside their phenotypic consequences and introduce base editor screen analysis with activity normalization (BEAN), a Bayesian network that uses per-guide editing outcomes provided by the reporter and target site chromatin accessibility to estimate variant impacts.

DNA-Diffusion: Leveraging Generative Models for Controlling Chromatin Accessibility and Gene Expression via Synthetic Regulatory Elements.

The challenge of systematically modifying and optimizing regulatory elements for precise gene expression control is central to modern genomics and synthetic biology. Advancements in generative AI have paved the way for designing synthetic sequences with the aim of safely and accurately modulating gene expression. We leverage diffusion models to design context-specific DNA regulatory sequences, which hold significant potential toward enabling novel therapeutic applications requiring precise modulation of gene expression.

Statistical inference with a manifold-constrained RNA velocity model uncovers cell cycle speed modulations.

Across a range of biological processes, cells undergo coordinated changes in gene expression, resulting in transcriptome dynamics that unfold within a low-dimensional manifold. Single-cell RNA-sequencing (scRNA-seq) only measures temporal snapshots of gene expression. However, information on the underlying low-dimensional dynamics can be extracted using RNA velocity, which models unspliced and spliced RNA abundances to estimate the rate of change of gene expression.

Benchmarking computational methods to identify spatially variable genes and peaks.

Spatially resolved transcriptomics offers unprecedented insight by enabling the profiling of gene expression within the intact spatial context of cells, effectively adding a new and essential dimension to data interpretation. To efficiently detect spatial structure of interest, an essential step in analyzing such data involves identifying spatially variable genes. Despite researchers having developed several computational methods to accomplish this task, the lack of a comprehensive benchmark evaluating their performance remains a considerable gap in the field.

Joint genotypic and phenotypic outcome modeling improves base editing variant effect quantification.

CRISPR base editing screens are powerful tools for studying disease-associated variants at scale. However, the efficiency and precision of base editing perturbations vary, confounding the assessment of variant-induced phenotypic effects. Here, we provide an integrated pipeline that improves the estimation of variant impact in base editing screens.

Dictys: dynamic gene regulatory network dissects developmental continuum with single-cell multiomics.

Gene regulatory networks (GRNs) are key determinants of cell function and identity and are dynamically rewired during development and disease. Despite decades of advancement, challenges remain in GRN inference, including dynamic rewiring, causal inference, feedback loop modeling and context specificity. To address these challenges, we develop Dictys, a dynamic GRN inference and analysis method that leverages multiomic single-cell assays of chromatin accessibility and gene expression, context-specific transcription factor footprinting, stochastic process network and efficient probabilistic modeling of single-cell RNA-sequencing read counts.

Integrative dissection of gene regulatory elements at base resolution.

Although vast numbers of putative gene regulatory elements have been cataloged, the sequence motifs and individual bases that underlie their functions remain largely unknown. Here, we combine epigenetic perturbations, base editing, and deep learning to dissect regulatory sequences within the exemplar immune locus encoding CD69. We converge on a ∼170 base interval within a differentially accessible and acetylated enhancer critical for CD69 induction in stimulated Jurkat T cells.