Biology subjects

Villoutreix, P.

Publications and source records attributed to Villoutreix, P..

10 recordsLinked to original sources

bioRxiv2026

Kontext: Kernel-based method bridging spatial domain identification and cell-cell communication

The spatial organization of biological tissues is characterized by spatial domains: regions with coherent transcriptional and anatomical properties, heavily influenced by local cell-cell interactions. Although identifying spatial domains is a common task in spatial transcriptomics, most existing methods provide limited mechanistic insight into their organization. We present Kontext, a kernel-based framework that jointly represents cells and their local environments, enabling the identification of spatial domains grounded in their underlying ligand-receptor interactions. This representation further provides complementary outputs for characterizing cellular microenvironments, spatial niches, and sender-receiver communication patterns. Kontext outperforms state-of-the-art methods in recovering reference domain annotations across several spatial transcriptomics technologies and biological models, while offering improved computational efficiency and scalability. Applying Kontext on a Xenium temporal mouse heart cryoinjury dataset, we characterize the spatiotemporal evolution of the injury and identify an undescribed sustained activity of the complement system in the communication between fibroblasts and macrophages during scar maturation. In a million-cell three-dimensional Open-ST dataset of human metastatic lymph node tissue, we combine spatial domain identification and niche analysis to define the boundary region between the tumor and its microenvironment, revealing distinct communication patterns between this interface and the tumor core.

bioinformatics↗

bioRxiv2026

Differential Analysis of Gene Spatial Organisation with Minkowski Functionals and Tensors

Spatial transcriptomics measures gene expression together with transcript coordinates in tissues. To date, comparing spatial gene expression patterns within and across samples remains challenging. We present here minkiPy, a geometric framework that computes, for each gene, a compact profile of morphological and topological descriptors based on Minkowski functionals and tensors. These profiles are defined in a shared feature space, enabling direct comparison of spatial organisation across genes, samples, and conditions, and the ranking of genes by the magnitude of their spatial reorganisation. We applied minkiPy to a MERFISH dataset of control and facioscapulohumeral muscular dystrophy myoblast cultures and to a Visium HD dataset of colorectal cancer and normal adjacent tissues, illustrating its utility across tissue types and spatial transcriptomics platforms. minkiPy is an open-source Python library available at https://github.com/BAUDOTlab/minkiPy.

bioinformatics↗

bioRxiv2025

jsPCA: fast, scalable, and interpretable identification of spatial domains and variable genes across multi-slice and multi-sample spatial transcriptomics data

Spatial transcriptomics technologies record genome-wide measurements of gene expression with high spatial resolution. These technologies generate large and high-dimensional datasets requiring efficient automated methods for their analysis. We introduce joint spatial PCA (jsPCA), a novel, fast, scalable and interpretable method for the automatic identification of spatial domains and variable genes in multi-slice and multi-sample spatial transcriptomics data. jsPCA relies on a simple mathematical formulation of a spatial covariance defined as the product of the gene expression covariance with the spatial autocorrelation. The principal components of this spatial covariance yield a biologically meaningful low-dimensional representation. From this representation, spatial domains are derived by simple clustering and spatially variable genes are identified directly from the principal component coefficients. A joint representation of multiple slices and samples without spatial alignment is obtained by computing common principal components via joint diagonalization. By leveraging data sparsity and non-convex manifold optimization, jsPCA leads to computing time in the order of seconds to minutes, substantially outperforming state-of-the-art approaches. We benchmarked jsPCA against 10 state-of-the-art methods on two reference databases. Our approach demonstrated excellent performance, comparable or better than state-of-the-art methods, while being much faster, interpretable, and scalable to very large datasets.

bioinformatics↗

bioRxiv2024

Spatial reconstruction of single-cell enhancer activity in a multicellular organism

Enhancers play an essential role in developmental processes by orchestrating the spatial and temporal regulation of gene expression. However, mapping the location of these regulatory elements in the genome and precisely characterizing their spatial and temporal activity remain important challenges. Here we introduce a novel in vivo and in silico method for spatial single-cell enhancer-reporter assays (spatial-scERA) designed to reconstruct the spatial activity of multiple candidate enhancer regions in parallel in a multicellular organism. Spatial-scERA integrates massively parallel reporter assays coupled with single-cell RNA sequencing (scRNA-seq) and spatial reconstruction using optimal transport, to map cell-type-specific enhancer activity at the single-cell level on a 3D virtual representation of the sample. We evaluated spatial-scERA in stage 6 Drosophila embryos using 25 candidate enhancers (including 19 uncharacterized regions), and validated the robustness of our predicted reconstructions by comparing them to microscopy images generated by in situ hybridization. Remarkably, spatial-scERA faithfully reconstructed the spatial activity of these enhancers, even when the enhancer-reporter construct was expressed in as few as 10 cells. Our results demonstrate the importance of integrating transcriptomic and spatial data for the accurate prediction of enhancer activity patterns in complex multicellular samples. Indeed, we found that chromatin modifications and open chromatin regions are often poor predictors of enhancer activity. Moreover, spatial data can often be essential for the accurate annotation of scRNA-seq clusters. Finally, we showed that spatial-scERA could be a powerful tool to link enhancers with their potential target genes. Overall, spatial-scERA provides a scalable approach to map spatio-temporal enhancer activity at single-cell resolution without the need for imaging or a priori knowledge of embryology and can be applied to any multicellular organism amenable to transgenesis.

genomics↗

bioRxiv2024

Assessing the Relative Contributions of Mosaic and Regulatory Developmental Modes from Single-Cell Trajectories

Development is a highly complex process consisting of coordinated cell proliferation, cell differentiation and spatial organization. Classically, two ways to specify cell types during development are hypothesized : mosaic and regulative modes. In the mosaic mode, a particular cell isolated from the rest of the embryo will nevertheless give rise to cells with a fate identical to the ones expected in normal development, thus relying on lineage-inherited factors. In the regulative mode, the fate of a cell depends on its interactions with its environment, and thus relies on space-dependant factors. Both modes often coexist in the development of a given animal. We propose to quantify their respective contributions from single-cell trajectories. C. elegans development provides a unique opportunity to elaborate such an approach. Indeed, its invariant lineage enables the integration of spatial positions, lineage relationships and protein expression data. Using the single cell protein expression profile as a readout of the cell state, we relate the contributions of the mosaic and the regulative modes to the following measurable quantities. The contribution of the mosaic mode, or lineage-inherited contribution is quantified by the strength of the relationship between the cell-cell lineage distance and the cell-cell expression distance. Similarly, the contribution of the regulative mode, or context-dependent contribution is quantified by the strength of the relationship between the cell-cell context distance and the cell-cell expression distance. The cell-cell context distance measures the similarity between the spatial neighborhoods of two cells based on the gene expression profiles of their neighbours. We assess the significance of these contributions by comparing the empirical results obtained on C. elegans data to artificial models generated using simple rules. With these measures, we show the co-existence of mosaic and regulative modes in the development of C. elegans. The relative contribution of these two modes varies across the different tissues and in time. In particular, we see in the skin tissue that during early development, the mosaic mode dominates while at later stages, regulative mode dominates, suggesting a convergence of single cell trajectories. These measures are general and can be applied to other datasets that will be made available with the progress of spatial transcriptomics and lineage-tracing, paving the way for a quantitative, unbiased and perturbation-free study of fundamental concepts in developmental biology.

systems biology↗

bioRxiv2024

MOTL: enhancing multi-omics matrix factorization with transfer learning

Joint matrix factorization is popular for extracting lower dimensional representations of multi-omics data but loses effectiveness with limited samples. Addressing this limitation, we introduce MOTL (Multi-Omics Transfer Learning), a framework that enhances MOFA (Multi-Omics Factor Analysis) by inferring latent factors for small multi-omics target datasets with respect to those inferred from a large heterogeneous learning dataset. We evaluated MOTL by designing simulated and real data protocols and demonstrated that MOTL improves the factorization of limited-sample multi-omics datasets when compared to factorization without transfer learning. When applied to actual glioblastoma samples, MOTL enhanced delineation of cancer status and subtype.

bioinformatics↗

bioRxiv2023

Bone elongation in the embryo occurs without column formation in the growth plate

Chondrocyte columns, which are a hallmark of growth plate architecture, play a central role in bone elongation. Columns are formed by clonal expansion following rotation of the division plane, resulting in a stack of cells oriented parallel to the growth direction. However, despite decades of research, column structure has thus far been studied only in two dimensions. To fill this knowledge gap, we analyzed hundreds of Confetti multicolor clones in growth plates of mouse embryos using a pipeline comprising 3D imaging and algorithms for morphometric analysis. Surprisingly, analysis of the elevation angles between neighboring pairs of cells revealed that most cells did not display the typical stacking pattern associated with column formation, implying incomplete rotation of the division plane. Morphological analysis revealed that although embryonic clones were elongated, they formed clusters oriented perpendicular to the growth direction. Analysis of growth plates of postnatal mice revealed both complex columns, composed of both ordered and disordered cell stacks, and small, disorganized clusters located in the outer edges. Our finding that embryonic growth plates function without forming columns suggests that longitudinal bone growth is regulated by different cellular mechanisms during pre- and postnatal development. Moreover, the observed complex columnar and cluster arrangements may serve other, yet unknown morphogenetic functions. More broadly, our findings provide a new understanding of the cellular mechanisms underlying growth plate activity and bone elongation during development.

developmental biology↗

bioRxiv2020

3D MAPs discovers the morphological sequence chondrocytes undergo in the growth plate and the regulatory role of GDF5 in this process

The activity of the epiphyseal growth plates, which drive longitudinal growth of long bones, is dependent on the ability of chondrocytes to change their shape and size extensively as they differentiate. However, organ size, extracellular matrix density and cell number have hindered the study of chondrocyte morphology. Here, we describe a new pipeline called 3D Morphometric Analysis for Phenotypic significance (3D MAPs), which overcomes these obstacles. By using 3D MAPs, we have created an image database of hundreds of thousands of cells from orthologous long bones. Analysis of this database revealed the growth strategies that chondrocytes use during differentiation. We found that chondrocytes employed both allometric and isometric growth, and that allometric growth is achieved by changes either in volume or surface area along a specific cell axis in a zone-specific manner. Additionally, we discovered a new organization of chondrocytes within the growth plate, where cells are orientated such that their longest axis always aligns with the dorsal-ventral axis of the bone. To demonstrate the ability of 3D MAPs to explore mechanisms of growth plate regulation, we studied the abnormally short tibiae of Gdf5-null mice. 3D MAPs identified aberrant cellular growth behaviors which resulted in a 3-fold reduction in volumetric cell growth, as well as affected cell morphology and orientation, highlighting GDF5 as a new regulator of growth plate activity. Overall, our findings provide new insight into the morphological sequence that chondrocytes undergo during differentiation and highlight the ability of 3D MAPs to uncover molecular and cellular mechanisms regulating this process. More broadly, this work provides a new framework for studying growth plate biology.

developmental biology↗

bioRxiv2018

Multiscale analysis of 3D nuclear morphology reveals new insights into growth plate organization in mice

The shape of the nucleus is tightly associated with cell morphology, the mechanical environment, and differentiation and transcriptional states. Yet, imaging of nuclei in three dimensions while preserving the spatial context of the tissue has been highly challenging. Here, using the embryonic tibial growth plate as a model for cell differentiation, we study nuclear morphology by imaging cleared samples by light-sheet fluorescence microscopy. Next, we quickly segmented tens of thousands of nuclei using several open-source tools including machine learning. Finally, segmented nuclei underwent morphometric analysis and 3D spatial reconstruction using newly designed algorithms. Our method revealed differences in nuclear morphology between chondrocytes at different differentiation stages. Additionally, we identified different morphological patterns in opposing growth plates, such as gradients of volume and surface area, as well as features characteristic of specific growth plate zones, such as sphericity and orientation. Altogether, this work supports a link between nuclear morphology and cell differentiation. Moreover, it demonstrates the suitability of our approach for studying the relationships between nuclear morphology and organ development.\n\nAuthor summaryThere has been a growing interest in the relationship between nuclear morphology and its regulation of gene expression. However, to study global patterns of nuclear morphology within a tissue we must address the problem of acquiring and analyzing multiscale data, ranging from the tissue level through to subcellular resolution. We have established a new pipeline that enables acquisition and segmentation of hundreds of thousands of nuclei at a resolution that allows quantitative analysis. Moreover we have developed new algorithms that allow superimposing morphological aspects of hundreds of thousands of nuclei onto a visual representation of the entire tissue, allowing us to study nuclear morphology at an organ level. Using mouse growth plates as a model for the relationship between nuclear morphology and tissue differentiation, we show that nuclei change different aspects of their morphology during chondrocyte differentiation. Growth plates are usually described generically in the literature, suggesting they lack unique characteristics. We challenge this dogma by showing that morphological features such as volume distribute differently in opposing growth plates. Altogether, this work highlights the possible role of nuclear shape in the regulation of cell differentiation and demonstrates that our approach enables the study of nuclear morphology patterns within a tissue.

developmental biology↗

bioRxiv2017

Synthesizing developmental trajectories

Dynamical processes in biology are studied using an ever-increasing number of techniques, each of which brings out unique features of the system. One of the current challenges is to develop systematic approaches for fusing heterogeneous datasets into an integrated view of multivariable dynamics. We demonstrate that heterogeneous data fusion can be successfully implemented within a semi-supervised learning framework that exploits the intrinsic geometry of high-dimensional datasets. We illustrate our approach using a dataset from studies of pattern formation in Drosophila. The result is a continuous trajectory that reveals the joint dynamics of gene expression, subcellular protein localization, protein phosphorylation, and tissue morphogenesis. Our approach can be readily adapted to other imaging modalities and forms a starting point for further steps of data analytics and modeling of biological dynamics.

developmental biology↗