bioRxiv Science⌕ Search

Biology subjects

Christiaens, V.

Publications and source records attributed to Christiaens, V..

17 recordsLinked to original sources
bioRxiv2026

System-wide extraction of cis-regulatory rules from sequence-to-function models in human neural development

The genomic cis-regulatory code (CRC) underlies spatiotemporal specificity of gene expression. While sequence-to-function (S2F) models can accurately encode the CRC of transcriptional enhancers, decoding these models into human-interpretable rules remains a major challenge. Here we tackle this challenge in human neural development, for which we generate two new single-cell multiome atlases, one from a human embryo and one from neural tube organoids. We use this comparative framework to robustly extract combinations of transcription factor (TF) binding sites that are necessary and sufficient to design enhancers. As such we extract cis-regulatory rules for dorsal-ventral progenitors, neural crest, mesenchyme and neurons. To enable this, we develop a new strategy and computational package, called TF-MINDI, to embed, cluster, and annotate candidate TF binding sites, and to extract combinatorial rules for each cell type. We evaluate rule-based models in conjunction with blackbox S2F models through simulations, evolutionary comparisons with zebrafish, topic modeling, and enhancer reporter-assays. Our findings show robust and interpretable rule extraction and constitute a step forward in deciphering, explaining, and formalizing the CRC. TF-MINDI is available at: https://github.com/aertslab/TF-MINDI.

bioinformatics↗
bioRxiv2025

HyDrop v2: Scalable atlas construction for training sequence-to-function models

Deciphering cis-regulatory logic underlying cell type identity is a fundamental question in biology. Single-cell chromatin accessibility (scATAC-seq) data has enabled training of sequence-to-function deep learning models allowing decoding of enhancer logic and design of synthetic enhancers. Training such models requires large amounts of high-quality training data across species, organs, development, aging, and disease. To facilitate the cost-effective generation of large scATAC-seq atlases for model training, we developed a new version of the open-source microfluidic system HyDrop with increased sensitivity and scale: HyDrop v2. We generated HyDrop v2 atlases for the mouse cortex and Drosophila embryo development and compared them to atlases generated on commercial platforms. HyDrop v2 data integrates seamlessly with commercially available chromatin accessibility methods (10x Genomics). Differentially accessible regions and motif enrichment across cell types are equivalent between HyDrop-v2 and 10x atlases. Sequence-to-function models trained on either atlas are comparable as well in terms of enhancer predictions, sequence explainability, and transcription factor footprinting. By offering accessible data generation, enhancer models trained on HyDrop-v2 and mixed atlases can contribute to unraveling cell-type specific regulatory elements in health and disease.

bioinformatics↗
bioRxiv2025

CREsted: modeling genomic and synthetic cell type-specific enhancers across tissues and species

Sequence-based deep learning models have become the state of the art for the analysis of the genomic regulatory code. Particularly for transcriptional enhancers, deep learning models excel at deciphering sequence features and grammar that underlie their spatiotemporal activity. To enable end-to-end enhancer modeling and design, we developed a software and modeling package, called CREsted. It combines preprocessing starting from single-cell ATAC-seq data; modeling with a choice of several architectures for training classification and regression models on either topics or pseudobulk peak heights; sequence design using multiple strategies; and downstream analysis through a collection of tools to locate transcription factor (TF) binding sites, infer the effect of a TF (activating or repressing) on enhancer accessibility, decipher enhancer grammar, and score gene loci. We demonstrate CREsted using a mouse cortex model that we validate using the BICCN collection of in vivo validated mouse brain enhancers. Classical enhancers in immune cells, including the IFNB1 enhanceosome are revisited using a PBMC model, and we assess the accuracy of TF binding site predictions with ChIP-seq. Additionally, we use CREsted to compare mesenchymal-like cancer cell states between tumor types; and we investigate different fine-tuning strategies of Borzoi within CREsted, comparing their performance and explainability with CREsted models trained from scratch. Finally, we train a CREsted model on a scATAC-seq atlas of zebrafish development and use this to design and in vivo validate cell type-specific synthetic enhancers in three tissues. For varying datasets, we demonstrate that CREsted facilitates efficient training and analyses, enabling scrutinization of the enhancer logic and design of synthetic enhancers across tissues and species. CREsted is available at https://crested.readthedocs.io.

genomics↗
bioRxiv2024

Enhancer-driven cell type comparison reveals similarities between the mammalian and bird pallium

Combinations of transcription factors govern the identity of cell types, which is reflected by enhancer codes in cis-regulatory genomic regions. Cell type-specific enhancer codes at nucleotide-level resolution have not yet been characterized for the mammalian neocortex. It is currently unknown whether these codes are conserved in other vertebrate brains, and whether they are informative to resolve homology relationships for species that lack a neocortex such as birds. To compare enhancer codes of cell types from the mammalian neocortex with those from the bird pallium, we generated single-cell multiome and spatially-resolved transcriptomics data of the chicken telencephalon. We then trained deep learning models to characterize cell type-specific enhancer codes for the human, mouse, and chicken telencephalon. We devised three metrics that exploit enhancer codes to compare cell types between species. Based on these metrics, non-neuronal and GABAergic cell types show a high degree of regulatory similarity across vertebrates. Proposed homologies between mammalian neocortical and avian pallial excitatory neurons are still debated. Our enhancer code based comparison shows that excitatory neurons of the mammalian neocortex and the avian pallium exhibit a higher degree of divergence than other cell types. In contrast to existing evolutionary models, the mammalian deep layer excitatory neurons are most similar to mesopallial neurons; and mammalian upper layer neurons to hyper- and nidopallial neurons based on their enhancer codes. In addition to characterizing the enhancer codes in the mammalian and avian telencephalon, and revealing unexpected correspondences between cell types of the mammalian neocortex and the chicken pallium, we present generally applicable deep learning approaches to characterize and compare cell types across species via the genomic regulatory code.

genomics↗
bioRxiv2022

Enhancer grammar of liver cell types and hepatocyte zonation states

Cell type identity is encoded by gene regulatory networks (GRN), in which transcription factors (TFs) bind to enhancers to regulate target gene expression. In the mammalian liver, lineage TFs have been characterized for the main cell types, including hepatocytes. Hepatocytes cover a relatively broad cellular state space, as they differ significantly in their metabolic state, and function, depending on their position with respect to the central or portal vein in a liver lobule. It is unclear whether this spatially defined cellular state space, called zonation, is also governed by a well-defined gene regulatory code. To address this challenge, we have mapped enhancer-GRNs across liver cell types at high resolution, using a combination of single cell multiomics, spatial omics, GRN inference, and deep learning. We found that cell state changes in transcription and chromatin accessibility in hepatocytes, liver sinusoidal endothelial cells and hepatic stellate cells depend on zonation. Enhancer-GRN mapping suggests that zonation states in hepatocytes are driven by the repressors Tcf7l1 and Tbx3, that modulate the core hepatocyte GRN, controlled by Hnf4a, Cebpa, Hnf1a, Onecut1 and Foxa1, among others. To investigate how these TFs cooperate with cell type TFs, we performed an in vivo massively parallel reporter assay on 12,000 hepatocyte enhancers and used these data to train a hierarchical deep learning model (called DeepLiver) that exploits both enhancer accessibility and activity. DeepLiver confirms Cebpa, Onecut, Foxa1, Hnf1a and Hnf4a as drivers of enhancer specificity in hepatocytes; Tcf7l1/2 and Tbx3 as regulators of the zonation state; and Hnf4a, Hnf1a, AP-1 and Ets as activators. Finally, taking advantage of in silico mutagenesis predictions from DeepLiver and enhancer assays, we confirmed that the destruction of Tcf7l1/2 or Tbx3 motifs in zonated enhancers abrogates their zonation bias. Our study provides a multi-modal understanding of the regulatory code underlying hepatocyte identity and their zonation state, that can be exploited to engineer enhancers with specific activity levels and zonation patterns.

genomics↗
bioRxiv2022

SCENIC+: single-cell multiomic inference of enhancers and gene regulatory networks

Joint profiling of chromatin accessibility and gene expression of individual cells provides an opportunity to decipher enhancer-driven gene regulatory networks (eGRN). Here we present a new method for the inference of eGRNs, called SCENIC+. SCENIC+ predicts genomic enhancers along with candidate upstream transcription factors (TF) and links these enhancers to candidate target genes. Specific TFs for each cell type or cell state are predicted based on the concordance of TF binding site accessibility, TF expression, and target gene expression. To improve both recall and precision of TF identification, we curated and clustered more than 40,000 position weight matrices that we could associate with 1,553 human TFs. We validated and benchmarked each of the SCENIC+ components on diverse data sets from different species, including human peripheral blood mononuclear cell types, ENCODE cell lines, human melanoma cell states, and Drosophila retinal development. Next, we exploit SCENIC+ predictions to study conserved TFs, enhancers, and GRNs between human and mouse cell types in the cerebral cortex. Finally, we provide new capabilities that exploit the inferred eGRNs to study the dynamics of gene regulation along differentiation trajectories; to map regulatory activities onto tissues using spatial omics data; and to predict the effect of TF perturbations on cell state. SCENIC+ provides critical insight into gene regulation, starting from multiome atlases of scATAC-seq and scRNA-seq. The SCENIC+ suite is available as a set of Python modules at https://scenicplus.readthedocs.io.

bioinformatics↗
bioRxiv2022

Cell type directed design of synthetic enhancers

Transcriptional enhancers act as docking stations for combinations of transcription factors and thereby regulate spatiotemporal activation of their target genes. A single enhancer, of a few hundred base pairs in length, can autonomously and independently of its location and orientation drive cell-type specific expression of a gene or transgene. It has been a long-standing goal in the field to decode the regulatory logic of an enhancer and to understand the details of how spatiotemporal gene expression is encoded in an enhancer sequence. Recently, deep learning models have yielded unprecedented insight into the enhancer code, and well-trained models are reaching a level of understanding that may be close to complete. As a consequence, we hypothesized that deep learning models can be used to guide the directed design of synthetic, cell type specific enhancers, and that this process would allow for a detailed tracing of all enhancer features at nucleotide-level resolution. Here we implemented and compared three different design strategies, each built on a deep learning model: (1) directed sequence evolution; (2) directed iterative motif implanting; and (3) generative design. We evaluated the function of fully synthetic enhancers to specifically target Kenyon cells in the fruit fly brain using transgenic animals. We then exploited this concept further by creating "dual-code" enhancers that target two cell types, and minimal enhancers smaller than 50 base pairs that are fully functional. By examining the trajectories followed during state space searches towards functional enhancers, we could accurately define the enhancer code as the optimal strength, combination, and relative distance of TF activator motifs, and the absence of TF repressor motifs. Finally, we applied the same three strategies to successfully design human enhancers. In conclusion, enhancer design guided by deep learning leads to better understanding of how enhancers work and shows that their code can be exploited to manipulate cell states.

genomics↗
bioRxiv2022

Shared enhancer gene regulatory networks between wound and oncogenic programs

Wound response programs are often activated during neoplastic growth in tumors. In both wound repair and tumor growth, cells respond to acute stress and balance the activation of multiple programs including apoptosis, proliferation, and cell migration. Central to those responses are the activation of the JNK/MAPK and JAK/STAT signaling pathways. Yet, to what extent these signaling cascades interact at the cis-regulatory level, and how they orchestrate different regulatory and phenotypic responses is still unclear. Here, we aim to characterize the regulatory states that emerge and cooperate in the wound response, using the Drosophila melanogaster wing disc as a model system, and compare these with cancer cell states induced by rasV12scrib-/- in the eye disc. We used single-cell multiome profiling to derive enhancer Gene Regulatory Networks (eGRNs) by integrating chromatin accessibility and gene expression signals. We identify a "proliferative" eGRN, active in the majority of wounded cells and controlled by AP-1 and STAT. In a smaller, but distinct population of wound cells, a "senescent" eGRN is activated and driven by C/EBP-like transcription factors (Irbp18, Xrp1, Slow border, and Vrille) and Scalloped. In tumor cells on the other hand, both eGRN signatures are simultaneously active within the same cell. Our single-cell multiome and eGRNs resource offers an in-depth characterisation of the senescence markers, together with a new perspective on the shared gene regulatory programs acting during wound response and oncogenesis.

cell biology↗
bioRxiv2021

Decoding gene regulation in the fly brain

The Drosophila brain is a work horse in neuroscience. Single-cell transcriptome analysis 1-5, 3D morphological classification 6, and detailed EM mapping of the connectome 7-10 have revealed an immense diversity of neuronal and glial cell types that underlie the wide array of functional and behavioral traits in the fruit fly. The identities of these cell types are controlled by - still unknown - gene regulatory networks (GRNs), involving combinations of transcription factors that bind to genomic enhancers to regulate their target genes. To characterize the GRN for each cell type in the Drosophila brain, we profiled chromatin accessibility of 240,919 single cells spanning nine developmental timepoints, and integrated this data with single-cell transcriptomes. We identify more than 95,000 regulatory regions that are used in different neuronal cell types, of which around 70,000 are linked to specific developmental trajectories, involving neurogenesis, reprogramming and maturation. For 40 cell types, their uniquely accessible regions could be associated with their expressed transcription factors and downstream target genes, through a combination of motif discovery, network inference techniques, and deep learning. We illustrate how these "enhancer-GRNs" can be used to reveal enhancer architectures leading to a better understanding of neuronal regulatory diversity. Finally, our atlas of regulatory elements can be used to design genetic driver lines for specific cell types at specific timepoints, facilitating the characterization of brain cell types and the manipulation of brain function.

genomics↗
bioRxiv2021

Analysis of long and short enhancers in melanoma cell states

Understanding how enhancers drive cell type specificity and efficiently identifying them is essential for the development of innovative therapeutic strategies. In melanoma, the melanocytic (MEL) and the mesenchymal-like (MES) states present themselves with different responses to therapy, making the identification of specific enhancers highly relevant. Using massively parallel reporter assay (MPRA) in a panel of patient-derived melanoma lines (MM lines), we set to identify and decipher melanoma enhancers by first focusing on regions with state specific H3K27 acetylation close to differentially expressed genes. A more in-depth evaluation of those regions was then pursued by investigating the activity of ATAC-seq peaks found therein along with a full tiling of the acetylated regions with 190 bp sequences. Activity was observed in more than 60% of the selected regions and we were able to precisely locate the active regions within ATAC-seq peaks. Comparison of sequence content with activity, using the deep learning model DeepMEL2, revealed that AP-1 alone is responsible for the MES enhancer activity, while SOX and MITF both influence MEL enhancer activity with SOX being required to achieve high levels of activity. Overall, our MPRA assays shed light on the relationship between long and short sequences in terms of their sequence content, enhancer activity, and specificity as reporters across melanoma cell states.

genomics↗
bioRxiv2019

Cross-species analysis of melanoma enhancer logic using deep learning

Genomic enhancers form the central nodes of gene regulatory networks by harbouring combinations of transcription factor binding sites. Deciphering the combinatorial code by which these binding sites are assembled within enhancers is indispensable to understand their regulatory involvement in establishing a cells phenotype, especially within biological systems with dysregulated gene regulatory networks, such as melanoma. In order to unravel the enhancer logic of the two most common melanoma cell states, namely the melanocytic and mesenchymal-like state, we combined comparative epigenomics with machine learning. By profiling chromatin accessibility using ATAC-seq on a cohort of 27 melanoma cell lines across six different species, we demonstrate the conservation of the two main melanoma states and their underlying master regulators. To perform an in-depth analysis of the enhancer architecture, we trained a deep neural network, called DeepMEL, to classify melanoma enhancers not only in the human genome, but also in other species. DeepMEL revealed the presence, organisation and positional specificity of important transcription factor binding sites. Together, this extensive analysis of the melanoma enhancer code allowed us to propose the concept of a core regulatory complex binding to melanocytic enhancers, consisting of SOX10, TFAP2A, MITF and RUNX, and to disentangle their individual roles in regulating enhancer accessibility and activity.

genomics↗
bioRxiv2019

Identification of genomic enhancers through spatial integration of single-cell transcriptomics and epigenomics

Single-cell technologies allow measuring chromatin accessibility and gene expression in each cell, but jointly utilizing both layers to map bona fide gene regulatory networks and enhancers remains challenging. Here, we generate independent single-cell RNA-seq and single-cell ATAC-seq atlases of the Drosophila eye-antennal disc and spatially integrate the data using a virtual latent space that mimics the organization of the 2D tissue. To validate spatially predicted enhancers, we use a large collection of enhancer-reporter lines and identify [~]85% of enhancers in which chromatin accessibility and enhancer activity are coupled. Next, we infer enhancer-to-gene relationships in the virtual space, finding that genes are regulated by multiple redundant enhancers. Exploiting cell-type specific enhancers, we deconvolute cell-type specific effects of bulk-derived chromatin accessibility QTLs. Finally, we discover that Prospero drives neuronal differentiation through the binding of a GGG motif. In summary, we provide a comprehensive spatial characterization of gene regulation in a 2D tissue.

genomics↗
bioRxiv2019

Single-cell gene regulatory network analysis reveals new melanoma cell states and transition trajectories during phenotype switching

Melanoma is notorious for its cellular heterogeneity, which is at least partly due to its ability to transition between alternate cell states. Similarly to EMT, melanoma cells with a melanocytic phenotype can switch to a mesenchymal-like phenotype. However, scattered emerging evidence indicates that additional, intermediate state(s) may exist. In order to search for such new melanoma states and decipher their underlying gene regulatory network (GRN), we extensively studied ten patient-derived melanoma cultures by single-cell RNA-seq of >39,000 cells. Although each culture exhibited a unique transcriptome, we identified shared gene regulatory networks that underlie the extreme melanocytic and mesenchymal cell states, as well as one (stable) intermediate state. The intermediate state was corroborated by a distinct open chromatin landscape and governed by the transcription factors EGR3, NFATC2, and RXRG. Single-cell migration assays established that this \"transition\" state exhibits an intermediate migratory phenotype. Through a dense time-series sampling of single cells and dynamic GRN inference, we unraveled the sequential and recurrent arrangement of transcriptional programs at play during phenotype switching that ultimately lead to the mesenchymal cell state. We provide the scRNA-Seq data with 39,263 melanoma cells on our SCope platform and the ATAC-seq data on a UCSC hub to jointly serve as a resource for the melanoma field. Together, this exhaustive analysis of melanoma cell state diversity indicates that additional states exists between the two extreme melanocytic and mesenchymal-like states. The GRN we identified may serve as a new putative target to prevent the switch to mesenchymal cell state and thereby, acquisition of metastatic and drug resistant potential.

genomics↗
bioRxiv2018

Cis-topic modelling of single cell epigenomes

Single-cell epigenomics provides new opportunities to decipher genomic regulatory programs from heterogeneous samples and dynamic processes. We present a probabilistic framework called cisTopic, to simultaneously discover \"cis-regulatory topics\" and stable cell states from sparse single-cell epigenomics data. After benchmarking cisTopic on single-cell ATAC-seq data, single-cell DNA methylation data, and semi-simulated single-cell ChIP-seq data, we use cisTopic to predict regulatory programs in the human brain and validate these by aligning them with co-expression networks derived from single-cell RNA-seq data. Next, we performed a time-series single-cell ATAC-seq experiment after SOX10 perturbations in melanoma cultures, where cisTopic revealed dynamic regulatory topics driven by SOX10 and AP-1. Finally, machine learning and enhancer modelling approaches allowed to predict cell type specific SOX10 and SOX9 binding sites based on topic specific co-regulatory motifs. cisTopic is available as an R/Bioconductor package at http://github.com/aertslab/cistopic.

bioinformatics↗
bioRxiv2017

A single-cell catalogue of regulatory states in the ageing Drosophila brain

The diversity of cell types and regulatory states in the brain, and how these change during ageing, remains largely unknown. Here, we present a single-cell transcriptome catalogue of the entire adult Drosophila melanogaster brain sampled across its lifespan. Both neurons and glia age through a process of \"regulatory erosion\", characterized by a strong decline of RNA content, and accompanied by increasing transcriptional and chromatin noise. We identify more than 50 cell types by specific transcription factors and their downstream gene regulatory networks. In addition to neurotransmitter types and neuroblast lineages, we find a novel neuronal cell state driven by datilografo and prospero. This state relates to neuronal birth order, the metabolic profile, and the activity of a neuron. Our single-cell brain catalogue reveals extensive regulatory heterogeneity linked to ageing and brain function and will serve as a reference for future studies of genetic variation and disease mutations.

genomics↗
bioRxiv2017

Identification Of cis-Regulatory Mutations Generating De Novo Edges In Personalized Cancer Gene Regulatory Networks

The identification of functional non-coding mutations is a key challenge in the field of genomics, where whole-genome re-sequencing can swiftly generate a set of all genomic variants in a sample, such as a tumor biopsy. The size of the human regulatory landscape places a challenge on finding recurrent cis-regulatory mutations across samples of the same cancer type. Therefore, powerful computational approaches are required to sift through the tens of thousands of non-coding variants, to identify potentially functional variants that have an impact on the gene expression profile of the sample. Here we introduce an integrative analysis pipeline, called - cisTarget, to filter, annotate and prioritize non-coding variants based on their putative effect on the underlying 'personal' gene regulatory network. We first validate -cisTarget by re-analyzing three cases of oncogenic non-coding mutations, namely the TAL1 and LMO1 enhancer mutations in T-ALL, and the TERT promoter mutation in melanoma. Next, we re-sequenced the full genome of ten cancer cell lines of six different cancer types, and used matched transcriptome data and motif discovery to infer master regulators for each sample. We identified candidate functional non-coding mutations that generate de novo binding sites for these master regulators, and that result in the up-regulation of nearby oncogenic drivers. We finally validated the predictions using tertiary data including matched epigenome data. Our approach is generally applicable to re-sequenced cancer genomes, or other genomes, when a disease- or sample-specific gene signature is available for network inference. -cisTarget is available from http://mucistarget.aertslab.org.

genomics↗