Biology subjects

Schwartz, G. W.

Publications and source records attributed to Schwartz, G. W..

At least 19 recordsLinked to original sources

bioRxiv2026

Longitudinal whole transcriptomic profiling of live cells through domain adaptation

Tracking transcriptomic profiles of cells over time in response to developmental cues and environmental stimuli can reveal critical insights into the fundamental mechanisms of development and disease. However, longitudinal molecular profiling at the global transcriptome level remains a major challenge, as RNA sequencing fundamentally alters or destroys cells. To overcome these limitations, we developed PENNE, a deep-learning framework that infers whole-transcriptomic profiles directly from live-cell images. Using gated attention mechanisms, PENNE trains on spatial transcriptomic datasets to align morphological features with gene expression. To enable inferences from images, our model performs domain adaptation to eliminate discrepancies between stained and unstained tissue images, effectively transferring molecular information from tissue sections to live-cell imaging. PENNE accurately identifies cell-type-specific and radiation-response markers via imputed expression. Furthermore, using only live-cell images stained with a G2/M cell cycle marker, our model captures temporal gene dynamics, evidenced by strong correlations between predicted expression and both ground-truth cellular confluency and cell-cycle progression. By bridging the gap between data-rich spatial transcriptomics and the practicality of live-cell imaging, PENNE provides a powerful new framework for monitoring molecular temporal dynamics directly through morphological information. This approach enables a paradigm-shifting workflow, fusing transcriptome-wide data with live-cell microscopy to fuel the discovery of novel gene programs via scalable, non-invasive, real-time interrogation of cellular states.

bioinformatics↗

bioRxiv2026

ClumPyCells resolves spatial aggregation in complex tissues overcoming size biases

The spatial arrangement of cells within a tissue microenvironment shapes their interactions and cell states, which are essential for tissue development, homeostasis, and disease. Spatial -omics technologies can precisely map the location of each cell within complex tissue structures, while also profiling their protein content and transcriptional diversity. Various approaches have been developed to analyze spatial patterns of cell aggregation, repulsion, or random distribution within tissues. However, differences in cell morphology within a tissue can introduce significant bias. Cell size in particular is not accounted for and introduces challenges when quantifying the aggregation of cells or their molecular features. To overcome such limitations, we present ClumPyCells: a statistical framework that measures cell and marker aggregation within tissue while correcting for size morphology. ClumPyCells enables interpretation of cell aggregation, bypassing interfering cell types or tissue regions unrelated to the desired spatial correlation. We demonstrate the capabilities of ClumPyCells across several tumor types, including melanoma and colorectal cancer, and spatial -omics technologies such as spatial transcriptomics and proteomics, while benchmarking how cell-size differences contribute to misinterpretations. By correcting for disruptive cell types within a region of interest, ClumPyCells will determine new tissue patterns and structures without morphological interference.

bioinformatics↗

bioRxiv2025

SPAGHETTI leverages massive H\&E morphological models for phase contrast microscopy images with a generative deep learning approach

Phase contrast microscopy (PCM) is a powerful cell imaging method, one of the few technologies to delineate and track cell structure in live cells without staining. Despite PCMs great potential and popularity in monitoring live-cell populations, there is a lack of algorithms to extract morphological information from these images due to the lack of large training datasets. To overcome this challenge and enable advanced, high throughput quantitative analysis of PCM images, we introduce SPAGHETTI: a lightweight image translator built on a modified cycle-consistent generative adversarial network. SPAGHETTI translates PCM images into those resembling hematoxylin and eosin (H&E) pathological images which, due to the pervasive and widespread use in clinical settings, are the basis for most large-scale deep learning models for quantitative analyses. We demonstrate that by first using SPAGHETTI to translate PCM images into H&E-like images, we could achieve significantly improved performance on cell segmentation through the use of tissue and H&E-specific cell segmentation models. We also show that by passing translated PCM images across several independent datasets into H&E feature extractor models, we improve the performance of cell-type annotation, experimental media classification, and cell viability prediction. Overall, SPAGHETTI enables many quantitative analyses of PCM that were previously impossible and acts as a valuable preprocessing step to help researchers gather novel information about cell states through the downstream quantitative analysis of morphological features.

bioinformatics↗

bioRxiv2025

iSHARC: Integrating scMultiome data for heterogeneity and regulatory analysis in cancer

SummaryThe 10x Genomics single cell Multiome (scMultiome) assay enables the simultaneous profiling of chromatin accessibility and gene expression from the same nucleus, and has increasingly been utilized in revealing cellular heterogeneity and gene regulation in cancers. However, a dedicated bioinformatics pipeline specifically designed for this type of data is still lacking. Here we present iSHARC, a streamlined pipeline for quality control, modality integration, clustering, cell type annotation, and regulatory mechanism analysis of individual scMultiome data, as well as for integrating multiple samples. The main advantages of iSHARC are: 1) easy implementation, execution and extension through a modular Snakemake workflow management system; 2) flexible analysis and parameters customization via a single configuration file; and 3) comprehensive accessibility by providing different access points to results and detailed summary reports from a single run. Availability and implementationThis pipeline is an open-source software under the MIT license and it is freely available at https://github.com/yzeng-lol/iSHARC. Contactyong.zeng@uhn.ca or hansen.he@uhn.ca or mathieu.lupien@uhn.ca Supplementary informationSupplementary data are appended.

bioinformatics↗

bioRxiv2024

Enhancer-promoter hubs organize transcriptional networks promoting oncogenesis and drug resistance

Recent advances in high-resolution mapping of spatial interactions among regulatory elements support the existence of complex topological assemblies of enhancers and promoters known as enhancer-promoter hubs or cliques. Yet, organization principles of these multi-interacting enhancer-promoter hubs and their potential role in regulating gene expression in cancer remains unclear. Here, we systematically identified enhancer-promoter hubs in breast cancer, lymphoma, and leukemia. We found that highly interacting enhancer-promoter hubs form at key oncogenes and lineage-associated transcription factors potentially promoting oncogenesis of these diverse cancer types. Genomic and optical mapping of interactions among enhancer and promoter elements further showed that topological alterations in hubs coincide with transcriptional changes underlying acquired resistance to targeted therapy in T cell leukemia and B cell lymphoma. Together, our findings suggest that enhancer-promoter hubs are dynamic and heterogeneous topological assemblies with the potential to control gene expression circuits promoting oncogenesis and drug resistance.

molecular biology↗

bioRxiv2024

Visual Identification of Conspecifics Shapes Social Behavior in Mice

Recognizing conspecifics in order to determine how to interact with them appropriately is a fundamental goal of animal sensory systems. It has undergone selective pressure in nearly all species. Mice have a large repertoire of social behaviors that are the subject of a rapidly growing field of study in neuroscience. Mouse social interactions likely incorporate all available sensory modalities, and the vast majority of studies have not attempted to isolate them. Specifically the role of vision in mouse social interactions remains unclear. We developed a behavioral platform that allowed us to present a subject mouse the visual information of stimulus mice in isolation from olfactory, acoustic, and tactile cues. Our results indicate that the visual identification of the sex or individual identity of other mice influences behavior. These findings highlight the underappreciated role of vision in mouse social interactions and open new avenues to study the visual circuits underlying social behavior.

neuroscience↗

bioRxiv2024

NEST: Spatially-mapped cell-cell communication patterns using a deep learning-based attention mechanism

Dysregulation of communication between cells mediates complex diseases such as cancer and diabetes. However, detecting cell-cell communication (CCC) at scale remains one of the greatest challenges in transcriptomics. While gene expression measured with single-cell RNA sequencing and spatial transcriptomics reinvigorated computational approaches to detecting CCC, most existing methods exhibit high false positive rates, do not integrate spatial proximity of ligand-receptor interactions, and cannot detect CCC between individual cells. We overcome these challenges by presenting NEST (NEural network on Spatial Transcriptomics), which uses a graph attention network paired with an unsupervised contrastive learning approach to decipher patterns of communication while retaining the strength of each signal. We introduce new synthetic benchmarking experiments which demonstrate how NEST outperforms existing tools and detects biologically-relevant CCC along with directionality and confidence across spot- and cell-based technologies measuring several different tissues and diseases. In our applications, NEST identifies T-cell homing signals in human lymph nodes, aggressive cancer CCC in lung adenocarcinoma, and discovers new patterns of communication that act as relay networks in pancreatic cancer. Beyond two-dimensional data, we also highlight NESTs ability to detect CCC in three-dimensional spatial transcriptomic data.

bioinformatics↗

bioRxiv2023

A hematopoietic stem cell subset that retains memory of prior inflammatory stress accumulates in aging and clonal hematopoiesis

Inflammation activates many blood cell types, driving aging and malignancy. Yet, hematopoietic stem cells (HSCs) survive a lifetime of infection to sustain life-long blood production. To understand HSC adaptation to inflammation, we developed xenograft inflammation-recovery models and performed single cell multiomics on isolated human HSC. Two transcriptionally and epigenetically distinct HSC subsets expressing canonical HSC programs were identified. Only one showed sustained transcriptional and epigenetic changes after recovery from inflammatory treatments. This HSC inflammatory memory (HSC-iM) program is enriched in memory T cells and HSCs from recovered COVID-19 patients. Importantly, HSC-iM accumulates with age and with clonal hematopoiesis. Overall, heritable molecular alterations in a subset of human HSCs, an adaptation to long-term inflammatory stress, may predispose to heightened age-related risk of blood cancer and infection. One-Sentence SummaryInflammation across a lifetime rewires human HSCs to produce a distinct HSC subset with both beneficial and deleterious fitness consequences.

cell biology↗

bioRxiv2023

TooManyCellsInteractive: a visualization tool for dynamic exploration of single-cell data

As single-cell sequencing data sets grow in size, visualizations of large cellular populations become difficult to parse and require extensive processing to identify subpopulations of cells. Managing many of these charts is laborious for technical users and unintuitive for non-technical users. To address this issue, we developed TooManyCellsInteractive (TMCI), a browser-based JavaScript application for visualizing hierarchical cellular populations as an interactive radial tree. TMCI allows users to explore, filter, and manipulate hierarchical data structures through an intuitive interface while also enabling batch export of high-quality custom graphics. Here we describe the software architecture and illustrate how TMCI has identified unique survival pathways among drug-tolerant persister cells in a pan-cancer analysis. TMCI will help guide increasingly large data visualizations and facilitate multi-resolution data exploration in a user-friendly way.

bioinformatics↗

bioRxiv2023

CDK5 activity in retinal pigment epithelium contributes to gap junction dynamics during phagocytosis

Retinal pigment epithelium (RPE) at the back of the eye is a monolayer of cells with an extensive network of gap junctions that contributes to retinal health in a multitude of ways. One of those roles is the phagocytosis of photoreceptor outer segments. This renewal is under circadian regulation and peaks after light onset. Connexin 43 (Cx43) is the most predominantly expressed gap junction protein in RPE. In this study, we examine how gap junctions and specifically, Cx43 phosphorylation, contribute to phagocytosis in both human embryonic stem cell derived RPE and mouse RPE monolayers. We show that both Rac1 and CDK5 have differences in protein localization at different points in phagocytosis, and that by using their effectors, the capability of RPE for phagocytosis changes. CDK5 has not yet been reported in RPE tissue, and here we show that it likely regulates Cx43 localization and resulting electrical coupling. We find that gap junctions in RPE are temporally highly dynamic during phagocytosis and that regulation of gap junctions via phosphorylation is likely critical for maintaining eye health.

cell biology↗

bioRxiv2022

Mind the gap: decoding decreases in tonic firing in populations of spiking neurons

A stimulus can be encoded in a population of spiking neurons through any change in the statistics of the joint spike pattern, yet we commonly summarize single-trial population activity by the summed spike rate across cells: the population peri-stimulus time histogram (pPSTH). For neurons with low baseline spike rate that encode a stimulus with a rate increase, this simplified representation works well, but for populations with high baseline rates and heterogeneous response patterns, the pPSTH has limited utility in capturing the neural representation of the stimulus. We simulated populations of spiking neurons that varied in size, baseline rate, burst statistics, and correlation, and we measured how these populations represent decreases (gaps) in spike rate. We introduce a different representation of the population spike pattern which we call an "information train," and we show that it is more flexible and robust than the pPSTH in capturing stimulus information across different types of neuronal populations. In particular, we use this tool to study populations with varying levels of burstiness in their spiking statistics. We find that there is an optimal level of burstiness for gap detection that is robust to several other parameters of the population. Next, we consider this theoretical result in the context of experimental data from different types of retinal ganglion cells and determine that the baseline spike statistics of a particular, recently identified type support nearly optimal detection of both the onset and strength of a contrast step.

neuroscience↗

bioRxiv2021

Multiple time scales of adaptation allow cones to encode the inputs created by visual exploration of natural scenes

Primates explore their visual environment by making frequent saccades, discrete and ballistic eye movements that direct the fovea to specific regions of interest. Saccades produce large and rapid changes in input. The magnitude of these changes and the limited signaling range of visual neurons means that effective encoding requires rapid adaptation. Here, we explore how cone photoreceptors maintain sensitivity under these conditions. Adaptation makes cone responses to naturalistic stimuli highly nonlinear and dependent on stimulus history. Such responses cannot be explained by linear or linear-nonlinear models but are well explained by a biophysical model of phototransduction with fast and slow adaptational mechanisms. The resulting model can predict cone responses to a broad range of stimuli and enables the design of stimuli that elicit specific (e.g. linear) cone photocurrents. These advances will provide a foundation for investigating the contributions of cones and post-cone processing to visual function.

neuroscience↗

bioRxiv2021

Multiomics single-cell analysis of human pancreatic islets reveals novel cellular states in health and type 1 diabetes

Type 1 diabetes (T1D) is an autoimmune disease of only partially defined etiology in which immune cells destroy insulin-producing beta cells. Using single-cell transcriptomics and an advanced analytical strategy to assess pancreatic islets of T1D, autoantibody-positive, and non-diabetic organ donors, we identified both canonical cell types and rare insulin-expressing cells with a hybrid mixture of endocrine and exocrine gene signatures within all donors. We further found elevated expression of MHC Class II pathway genes in exocrine ductal cells of T1D donors, which we confirmed through CyTOF, in situ imaging mass cytometry, and immunofluorescence analysis. Taken together, our multimodal analyses identify novel cell types and processes that may contribute to T1D immunopathogenesis and provide new cellular and molecular insights into human pancreas function.

genomics↗

bioRxiv2020

An offset ON-OFF receptive field is created by gap junctions between distinct types of retinal ganglion cells

Receptive fields (RFs) are a foundational concept in sensory neuroscience. The RF of a sensory neuron is shaped by the properties of its synaptic inputs from connected neurons. In the early visual system, retinotopic maps define a strict relationship between the location of a cells dendrites and its RF location in visual space1-3. Retinal ganglion cells (RGCs), the output cells of the retina, form dendritic mosaics that tile retinal space and have corresponding RF mosaics that tile visual space1,2. The precise location of dendrites in some RGCs has been shown to predict their RF shape4. Previously described ON-OFF RGCs have aligned dendrites in ON and OFF synaptic layers, so the cells respond to increments and decrements of light at the same locations in visual space5-8. Here we report a systematic offset between the ON and OFF RFs of an RGC type. Surprisingly, this property does not come from offset ON and OFF dendrites but instead arises from electrical synapses with RGCs of a different type. This circuit represents a new channel for direct communication between ON and OFF RGCs. Using a multi-cell model, we find that offset ON-OFF RFs could improve the precision with which edge location is represented in an RGC population.

neuroscience↗

bioRxiv2020

The Olivary Pretectal Nucleus Receives Visual Input of High Spatial Resolution

In the mouse, retinal output is computed by over 40 distinct types of retinal ganglion cells (RGCs) (Baden et al., 2016). Determining which of these many RGC types project to a retinorecipient region is a key step in elucidating the role that region plays in visually-mediated behaviors. Combining retrograde viral tracing and single-cell electrophysiology, we identify the RGC types which project to the olivary pretectal nucleus (OPN), a major visual structure. We find that retinal input to the OPN consists of a variety of intrinsically-photosensitive and conventional RGC types, the latter a diverse set of mostly ON RGCs. Surprisingly, while the OPN is most associated with the pupillary light reflex (PLR) pathway, requiring information about absolute luminance, we show that the majority of the retinal input to the OPN is from single cell type which transmits information unrelated to luminance. This ON-transient RGC accounts for two-thirds of the input to the OPN, and responds to small objects across a wide range of speeds. This finding suggests a role for the OPN in visually-mediated behaviors beyond the PLR. Significance statementThe olivary pretectal nucleus is a midbrain structure which receives direct input from retinal ganglion cells (RGC), and modulates pupil diameter in response to changing absolute light level. In the present study, we combine viral tracing and electrophysiology to identify the RGC types which project to the OPN. Surprisingly, the majority of its input comes from a single type which does not encode absolute luminance, but instead responds to small objects across a wide range of speeds. These findings are consistent with a role for the OPN apart from pupil control and suggest future experiments to elucidate its full role in visually-mediated behavior.

neuroscience↗

bioRxiv2020

Identification of Retinal Ganglion Cell Types and Brain Nuclei expressing the transcription factor Brn3c/Pou4f3 using a Cre recombinase knock-in allele.

Members of the POU4F/Brn3 transcription factor family have an established role in the development of retinal ganglion cell types (RGCs), the projection sensory neuron conveying visual information from the mammalian eye to the brain. Our previous work using sparse random recombination of a conditional knock-in reporter allele expressing Alkaline Phosphatase (AP) and intersectional genetics had identified three types of Pou4f3/Brn3c positive (Brn3c+) RGCs. Here, we describe a novel Brn3cCre mouse allele generated by serial Dre to Cre recombination. We use this allele to explore the expression overlap of Brn3c with Brn3a and Brn3b and the dendritic arbor morphologies and visual stimulus properties of Brn3c+ RGC types. Furthermore, we explore Brn3c-expressing brain nuclei. Our analysis reveals a much larger number of Brn3c+ RGCs and more diverse set of RGC types than previously reported. The majority of RGCs having expressed Brn3c during development are still Brn3c positive in the adult, and all of them express Brn3a while only about half express Brn3b. Intersection of Brn3b and Brn3c expression highlights an area of increased RGC density, similar to an area centralis, corresponding to part of the binocular field of view of the mouse. Brn3c+ neurons and projections are present in multiple brain nuclei. Brn3c+ RGC projections can be detected in the Lateral Geniculate Nucleus (LGN), Pretectal Area (PTA) and Superior Colliculus (SC) but also in the thalamic reticular nucleus (TRN), a visual circuit station that was not previously described to receive retinal input. Most Brn3c+ neurons of the brain are confined to the pretectum and the dorsal midbrain. Amongst theses we identify a previously unknown Brn3c+ subdivision of the deep mesencephalic nucleus (DpMe). Thus, our newly generated allele provides novel biological insights into RGC type classification, brain connectivity and midbrain cytoarchitectonic, and opens the avenue for specific characterization and manipulation of these structures.

neuroscience↗

bioRxiv2019

TooManyCells identifies and visualizes relationships of single-cell clades

Transcriptional programs contribute to phenotypic and functional cell states. While elucidation of cell state heterogeneity and its role in biology and pathobiology has been advanced by studying single cell level measurements, the underlying assumptions of current analytical methods limit the identification and exploration of cell clades. Unlike other methods, which produce a single uni-layer partition of cells ignoring echelons of cell states, we present TooManyCells, a software consisting of a suite of graph-based tools for efficient, global, and unbiased identification and visualization of cell clades while maintaining and presenting the relationship between cell states. TooManyCells provides a set of tools based on a matrix-free efficient divisive hierarchical spectral clustering algorithm wholly different from the prevalent Louvain-based methods. BirchBeer, the visualization component of TooManyCells, introduces a new approach for single cell analysis that is built on a concept intentionally orthogonal to the widely used dimensionality reduction methods. Together, this suite of tools provide a paradigm shift in the analysis and interpretation of single cell data by enabling simultaneous comparisons of cell states at context-and application-dependent scales. A byproduct of this shift is the immediate detection and visualization of rare populations that outperforms previous algorithms as demonstrated by applying these tools to existing single cell RNA-seq data sets from various mouse organs.

bioinformatics↗

bioRxiv2018

Differential Integration of Transcriptome and Proteome Identifies Pan-cancer Prognostic Biomarkers

High-throughput analysis of the transcriptome and proteome individually are used to interrogate complex oncogenic processes in cancer. However, an outstanding challenge is how to combine these complementary, yet partially disparate data sources to accurately identify tumor-specific gene-programs and clinical biomarkers. Here, we introduce inteGREAT for robust and scalable differential integration of high-throughput measurements. With inteGREAT, each data source is represented as a co-expression network, which is analyzed to characterize the local and global structure of each node across networks. inteGREAT scores the degree by which the topology of each gene in both transcriptome and proteome networks are conserved within a tumor type, yet different from other normal or malignant cells. We demonstrated the high performance of inteGREAT based on several analyses: deconvolving synthetic networks, rediscovering known diagnostic biomarkers, establishing relationships between tumor lineages, and elucidating putative prognostic biomarkers which we experimentally validated. Furthermore, we introduce the application of a clumpiness measure to quantitatively describe tumor lineage similarity. Together, inteGREAT not only infers functional and clinical insights from the integration of transcriptomic and proteomic data sources in cancer, but also can be readily applied to other heterogeneous high-throughput data sources. inteGREAT is open source and available to download from https://github.com/faryabib/inteGREAT.

bioinformatics↗