Biology subjects

Zaritsky, A.

Publications and source records attributed to Zaritsky, A..

At least 19 recordsLinked to original sources

bioRxiv2026

Trustworthy in silico labeling via semantic visual interpretability of image-to-image translation

Cross-modality image translation promises to provide multiple layers of biological information from a single image input, yet its practical application is stalled by a lack of interpretability and the inability to account for model imperfections. In silico labeling, the inference of organelle localization from label-free images, is a primary example where this black-box nature limits adoption. We present Mask Interpreter, a generalized method for semantic visual interpretability of image-to-image translation models. By uncovering organelle-specific "explanation signatures", Mask Interpreter validates that models rely on authentic biological structures rather than spurious artifacts. Beyond biological validation, it outperforms traditional explainable AI (xAI) approaches, identifies batch effects and localizes prediction errors when ground-truth fluorescence is unavailable. Semantic confidence modeling further provides fine-grained reliability assessment at single-cell resolution, enabling the automated exclusion of artifacts from downstream analyses. By bridging the gap between computational inference and meaningful biological features, Mask Interpreter transforms in silico labeling into a reliable tool for scientific discovery across diverse biomedical imaging modalities.

cell biology↗

bioRxiv2026

DNA replication initiation causally sets the added mass for cell division in Escherichia coli

Faithful genome inheritance requires cell division to be precisely coordinated with DNA replication. The classical Cooper-Helmstetter paradigm posited that division occurs at a roughly constant time after replication initiation (initiation-to-division "ID-timer" model). This has since been challenged by conflicting "adder" phenomena, leaving the primary signal for division a fundamental enigma. Here, we resolve these discrepancies by establishing that cell division in Escherichia coli obeys a universal "ID-adder" rule: addition of a defined size increment from replication initiation. This rule is invariant across the physiological range of nutrition-affected growth rates, whereas other versions of the adder phenomenon are emergent properties shaped by inter-cycle correlations. Through population-level perturbations of initiation mass, division machinery, and replication termination, we demonstrate that replication initiation directly and causally sets the division schedule. This principle operates independently of replication termination and division protein levels, providing a robust mechanism for coordinating the genome duplication with the cell division cycle.

microbiology↗

bioRxiv2025

Spatially distinct chromatin compaction states predict neoadjuvant chemotherapy resistance in Triple Negative Breast Cancer

Organisation and dynamics of chromatin play a key role in regulation of cell state and function. In cancer, chromatin plasticity is known to be important in control of drug resistance, but the relationship between chromatin compaction and chemotherapy response within complex tissue settings remains unclear. Here, we measured single nuclei chromatin compaction using fluorescence lifetime imaging microscopy (FLIM) in situ in whole biopsies from 53 pre treatment and 14 post surgery tissue samples of human triple-negative breast cancer (TNBC) patients to determine whether single nuclei spatial chromatin compaction state can predict resistance to neoadjuvant chemotherapy (NACT). Bulk chromatin compaction across 53 pre-treatment core biopsies did not predict patient outcome. However, machine learning analysis revealed that a subset of patients exhibited distinct distributions of single nuclei with more open chromatin states, which was predictive of NACT-resistance. Graph neural network analysis established that the spatial arrangement of chromatin compaction contributed to prediction of NACT resistance and that chromatin compaction signatures were preserved in tissue state transitions from pre- to 14 post-NACT samples. Our findings shed new light on spatial control of chromatin structure and relationship to therapeutic resistance and establish a foundation for further molecular analysis of chromatin states in complex biological tissues.

systems biology↗

bioRxiv2025

Extrusion-modulated DnaA activity oscillations coordinate DNA replication with biomass growth

Robust control of DNA replication is fundamental to bacterial proliferation. In Escherichia coli, replication initiation is thought to be regulated by oscillations in DnaA activity, driven by DnaA-chromosome interactions that differ among leading models. However, direct evidence linking these oscillations to replication initiation has been lacking, and existing models fail to explain the observed decoupling of replication initiation from dnaA expression. Here, we establish a direct link between DnaA activity and replication initiation by demonstrating robust oscillations in DnaA activity, which peak precisely at replication initiation across diverse growth conditions and genetic perturbations. Notably, these oscillations persist even when dnaA transcription remains constant, suggesting a regulatory mechanism that modulates DnaA activity independently of its expression. Additionally, we propose an extrusion model in which DNA-binding proteins sense biomass-DNA imbalance and extrude DnaA from the chromosome to trigger replication, overcoming limitations of existing models. Consistent with this model, perturbation of the nucleoid-associated protein H-NS modulates DnaA activity and replication timing, supporting its mechanistic validity.

microbiology↗

bioRxiv2025

Context-dependent spatial multicellular network motifs for single-cell spatial biology

The clinical state of diseased tissue is caused by complex intercellular processes that go beyond pairwise cell-cell interactions and are difficult to infer due to the combinatorial explosion of such high-dimensionality. We present context-dependent identification of spatial motifs (CISM), a two-step method to identify local cell structures associated with a disease state in single cell spatial data. First, for each tissue, CISM enumerates structures of enriched reoccurring multicellular patterns that define modular motifs in the multicellular network. Second, discriminative motifs are selected according to the context - their presence in patients at different clinical disease states. By applying CISM, we show that modular structures composed of as little as 3-5 cells and their relative spatial arrangement can encode differences in clinical disease states in cohorts of triple-negative breast cancer (TNBC) and melanoma patients. Machine learning validation indicated that discriminative motifs outperform state-of-the-art methods for disease state prediction while enabling interpretation of which interactions in what spatial context are associated with these predictions. CISM-derived discriminative motifs may define an intermediate spatial scale of abstraction and modularity in multicellular organization and function with broad applicability in the domain of spatial single cell omics and beyond.

systems biology↗

bioRxiv2024

Cell-context dependent in silico organelle localization in label-free microscopy images

In silico labeling prediction of organelle fluorescence from label-free microscopy images has the potential to revolutionize our understanding of cells as integrated complex systems. However, out-of-distribution data caused by changes in the intracellular organization across cell types, cellular processes or perturbations, can lead to altered label-free images and impaired in silico labeling. We demonstrated that incorporating biological meaningful cell contexts, via a context-dependent model that we call CELTIC, enhanced in silico labeling prediction and enabled downstream analysis of out-of-distribution data such as cells undergoing mitosis, and cells located at the edge of the colony. These results suggest a link between cell context and intracellular organization. Using CELTIC to generate single cell images transitioning between different contexts enabled us to overcome inter-cell variability toward integrated characterization of organelles alterations in cellular organization. The explicit inclusion of context has the potential to harmonize multiple datasets, paving the way for generalized in silico labeling foundation models.

cell biology↗

bioRxiv2024

Spatiotemporal analysis of F-actin polymerization with micropillar arrays reveals synchronization between adhesion sites

We repurposed micropillar-arrays to quantify spatiotemporal inter-adhesion communication. Following the observation that integrin adhesions formed around pillar tops we relied on the precise repetitive spatial control of the pillars to reliably monitor F-actin dynamics in mouse embryonic fibroblasts as a model for spatiotemporal adhesion-related intracellular signaling. Using correlation-based analyses we revealed localized information-flows propagating between adjacent pillars that were integrated over space and time to synchronize the adhesion dynamics within the entire cell. Probing the mechanical regulation, we discovered that stiffer pillars or partial actomyosin contractility inhibition enhances inter-adhesion F-actin synchronization. Our results suggest that adhesions can communicate and highlight the potential of using micropillar arrays as a tool to measure spatiotemporal intracellular signaling.

cell biology↗

bioRxiv2024

Anomaly detection for high-content image-based phenotypic cell profiling

High-content image-based phenotypic profiling combines automated microscopy and analysis to identify phenotypic alterations in cell morphology and provide insight into the cells physiological state. Classical representations of the phenotypic profile can not capture the full underlying complexity in cell organization, while recent weakly machine-learning based representation-learning methods are hard to biologically interpret. We used the abundance of control wells to learn the in-distribution of control experiments and use it to formulate a self-supervised reconstruction anomaly-based representation that encodes the intricate morphological inter-feature dependencies while preserving the representation interpretability. The performance of our anomaly-based representations was evaluated for downstream tasks with respect to two classical representations across four public Cell Painting datasets. Anomaly-based representations improved reproducibility, Mechanism of Action classification, and complemented classical representations. Unsupervised explainability of autoencoder-based anomalies identified specific inter-feature dependencies causing anomalies. The general concept of anomaly-based representations can be adapted to other applications in cell biology.

bioinformatics↗

bioRxiv2024

Data-modeling the interplay between single cell shape, single cell protein expression, and tissue state

Changes in cell shape are fundamentally involved in signaling, intracellular organization, function, and intercellular interactions within tissues, in health and disease. Investigating the interplay between cell shape and protein expression was limited, until recently, by the number of proteins that can be imaged simultaneously or by population averaging. We combined spatial multiplexed single cell imaging and machine learning to systematically investigate the intricate relationships between cell shape and protein expression in the context of heterogeneous human cells in their native state in human tissue samples in situ. Our analysis established a universal bi-directional link between the cells shape and its protein expression across different cell types, diseases, and disease states in human tissues, enabling new applications. Machine learning interpretability showed that the contribution of shape features to a prediction can potentially infer new protein functions. Unbiased screening of the links between all pairs consisting of one protein and one cell type identified a subpopulation of large p53-positive tumor cells across two cancers. Ultimately, inclusion of single cell shape properties enhanced Graph Neural Network disease state prediction. Our results open the door to unraveling the intricate connections between protein expression at the single cell level, cell shape, tissue organization, and tissue state in a physiological context.

cell biology↗

bioRxiv2023

Formation of recurring transient Ca2+-based intercellular communities during Drosophila hematopoiesis

Tissue development occurs through a complex interplay between many individual cells. Yet, the fundamental question of how collective tissue behavior emerges from heterogeneous and noisy information processing and transfer at the single-cell level remains unknown. Here, we reveal that tissue scale signaling regulation can arise from local gap-junction mediated cell-cell signaling through the spatiotemporal establishment of an intermediate-scale of transient multicellular communication communities over the course of tissue development. We demonstrated this intermediate scale of emergent signaling using Ca2+ signaling in the intact, ex vivo cultured, live developing Drosophila hematopoietic organ, the Lymph Gland (LG). Recurrent activation of these transient signaling communities defined self-organized signaling "hotspots" that receive and transmit information to facilitate repetitive interactions with non-hotspot neighbors, transfer information across cells, and regulate the developmental progression of hotspots. Overall, this work bridges the scales between single-cell and emergent group behavior providing key mechanistic insight into how cells establish tissue-scale communication networks. Significance statementCells coordinate their internal state and behavior by exchanging information with other cells in their vicinity. These local interactions are integrated across space and time to enable synchronized function at the tissue scale. Using live microscopy imaging of the Drosophila Lymph Gland, and by applying computational analyses, we identified and characterized a new mode of cellular communication through self-organized recurring coordinated short-term activation at the intermediate scale of 3-8 cells, which we call "hotspots". We reveal that hotspots form over the course of tissue development, and are dependent on specific proteins, called gap-junctions, that enable communication between adjacent cells. Hotspots repeatedly transmit and retrieve information to and from their non-hotspot neighbors to spread information throughout the tissue to regulate and coordinate tissue function.

developmental biology↗

bioRxiv2023

Visual interpretability of image-based classification models by generative latent space disentanglement applied to in vitro fertilization

The success of deep learning in identifying complex patterns exceeding human intuition comes at the cost of interpretability. Non-linear entanglement of image features makes deep learning a "black box" lacking human meaningful explanations for the models decision. We present DISCOVER, a generative model designed to discover the underlying visual properties driving image-based classification models. DISCOVER learns disentangled latent representations, where each latent feature encodes a unique classification-driving visual property. This design enables "human-in-the-loop" interpretation by generating disentangled exaggerated counterfactual explanations. We apply DISCOVER to interpret classification of in-vitro fertilization embryo morphology quality. We quantitatively and systematically confirm the interpretation of known embryo properties, discover properties without previous explicit measurements, and quantitatively determine and empirically verify the classification decision of specific embryo instances. We show that DISCOVER provides human-interpretable understanding of "black-box" classification models, proposes hypotheses to decipher underlying biomedical mechanisms, and provides transparency for the classification of individual predictions.

systems biology↗

bioRxiv2023

Cell Width in Escherichia coli: Interrogating Coupling to Nucleoid Structure

Cell dimensions of rod-shaped bacteria such as Escherichia coli are connected to mass growth and chromosome replication. It divides {approx}20 min after termination of the replication cycle that initiates {approx}40 min earlier at a relatively constant mass. Cells enlarge by elongation only, but at faster growth in richer media they are also wider. Width determination occurs in the divisome during the division process, coupled, temporally and spatially, to the ratio between the rates of growth and replication. The elusive signal directing the mechanism for width determination is related to the tightly linked duplications of the nucleoid (DNA) and the sacculus (peptidoglycan), the only two structures (macro-molecules) existing in a single copy. Six biologically meaningful parameters related to the key number of replication positions are reasonable candidates to convey such a signal. The current analysis discovered that of these, nucleoid complexity is the most likely parameter affecting cell width. As a corollary, a new, indirect approach to estimate replication rate is revealed.

microbiology↗

bioRxiv2023

Machine learning inference of continuous single-cell state transitions during myoblast differentiation and fusion

Cells modifying their internal organization during continuous state-transitions, supporting functions from cell division to differentiation. However, tools to measure dynamic physiological states of individual transitioning cells are lacking. We combined live-cell imaging and machine learning to monitor ERK1/2-inhibited primary murine skeletal muscle precursor cells, that transition rapidly and robustly from proliferating myoblasts to post-mitotic myocytes and then fuse, forming multinucleated myotubes. Our model, trained using motility and actin intensity features from single cell tracking data, effectively tracked real-time continuous differentiation, revealing that differentiation occurs 7.5-14.5 hours post-induction, followed by fusion [~]3 hours later. Co-inhibition of ERK1/2 and p38 led to differentiation without fusion. Our model inferred co-inhibition leads to terminal differentiation, indicating that p38 is specifically required for transitioning from terminal differentiation to fusion. Our model also predicted that co-inhibition leads to changes in actin dynamics. Mass spectrometry supported these in silico predictions and suggested novel fusion and maturation regulators downstream of differentiation. Collectively, this approach can be adapted to various biological processes to uncover novel links between dynamic single-cell states and their functional outcomes.

cell biology↗

bioRxiv2022

Using unlabeled information of embryo siblings from the same cohort cycle to enhance in vitro fertilization implantation prediction

High content time-lapse embryo imaging assessed by machine learning is revolutionizing the field of in vitro fertilization (IVF). However, the vast majority of IVF embryos are not transferred to the uterus, and these masses of embryos with unknown implantation outcomes are ignored in current efforts that aim to predict implantation. Here, we explore whether, and to what extent the information encoded within "sibling" embryos from the same IVF cohort contribute to the performance of machine learning-based implantation prediction. First, we show that the implantation outcome is correlated with attributes derived from the cohort siblings. Second, we demonstrate that this unlabeled data boosts implantation prediction performance. Third, we characterize the cohort properties driving embryo prediction, especially those that rescued erroneous predictions. Our results suggest that predictive models for embryo implantation can benefit from the overlooked, widely available unlabeled data of sibling embryos by reducing inherent noise of the individual transferred embryo. Significance statementWe use in vitro fertilization (IVF) as a model to study the effect of genotypic and environmental variation on phenotype and demonstrate a potential translational application. This is achieved by associating the implantation potential of transferred embryos and the visual information encoded within their non-transferred "sibling" embryos from the same IVF cohort, and establishing that these cohort features contribute to consistent improvement in machine learning implantation prediction regardless of the embryo-focused model. Our results suggest a general concept where the uncertainty in the implantation potential for the transferred embryo can be reduced by information encapsulated in the correlated cohort embryos. Since the siblings data are routinely collected, incorporating cohort features in AI-driven embryo implantation prediction can have direct translational implications.

developmental biology↗

bioRxiv2020

Emergence of synchronized multicellular mechanosensing from spatiotemporal integration of heterogeneous single-cell information transfer

We quantitatively characterize how noisy and heterogeneous behaviors of individual cells are integrated across a population toward multicellular synchronization by studying the calcium dynamics in mechanically stimulated monolayers of endothelial cells. We used information-theory to quantify the asymmetric information-transfer between pairs of cells and define quantitative measures of how single cells receive or transmit information in the multicellular network. We find that cells take different roles in intercellular information-transfer and that this heterogeneity is associated with synchronization. Cells tended to maintain their roles between consecutive cycles of mechanical stimuli and reinforced them over time, suggesting the existence of a cellular "memory" in intercellular information transfer. Interestingly, we identified a subpopulation of cells characterized by higher probability of both receiving and transmitting information. These "communication hub" roles were stable - once a cell switched to a "communication hub" role it was less probable to switch to other roles. This stableness property of the cells led to gradual enrichment of communication hubs that was associated with the establishment of synchronization. Our analysis demonstrated that multicellular synchronization was established by effective information spread from the (local) single cell to the (global) group scale in the multicellular network. Altogether, we suggest that multicellular synchronization is driven by single cell communication properties, including heterogeneity, functional memory and information flow.

cell biology↗

bioRxiv2020

Quantifying the dynamics of long-range cell-cell mechanical communication

Cells sense, manipulate and respond to their mechanical microenvironment in a plethora of physiological processes, yet the understanding of how cells transmit, receive and interpret environmental cues to communicate with distant cells is severely limited due to lack of tools to quantitatively infer the complex tangle of dynamic cell-cell interactions in complicated environments. We present a computational method to systematically infer and quantify long-range cell-cell force transmission through the extracellular matrix (cell-ECM-cell communication) by correlating ECM remodeling fluctuations in between communicating cells and demonstrating that these fluctuations contain sufficient information to define unique signatures that robustly distinguish between different pairs of communicating cells. We demonstrate our method with finite element simulations and live 3D imaging of fibroblasts and cancer cells embedded in fibrin gels. While previous studies relied on the formation of a visible fibrous band extending between cells to inform on mechanical communication, our method detected mechanical propagation even in cases where visible bands never formed. We revealed that while contractility is required, band formation is not necessary, for cell-ECM-cell communication, and that mechanical signals propagate from one cell to another even upon massive reduction in their contractility. Our method sets the stage to measure the fundamental aspects of intercellular long-range mechanical communication in physiological contexts and may provide a new functional readout for high content 3D image-based screening. The ability to infer cell-ECM-cell communication using standard confocal microscopy holds the promise for wide use and democratizing the method.

cell biology↗

bioRxiv2020

Cytokinetic abscission is part of the mid-blastula transition switch in early zebrafish embryogenesis

Animal cytokinesis ends with the formation of a thin intercellular membrane bridge connecting the two newly formed sibling cells that is ultimately resolved by abscission. While mitosis is completed within 15 minutes, the intercellular bridge can persist for hours, maintaining a physical connection between sibling cells and allowing exchange of cytosolic components. Although cell-cell communication is fundamental for development, the potential role of intercellular bridges during embryogenesis have not been fully elucidated. Here, we found that in early zebrafish (Danio rerio) embryogenesis, abscission is delayed and cells do not resolve their intercellular bridges until midblastula transition (MBT), giving rise to the formation of small inter-connected cell clusters. Interestingly, abscission commences during the MBT switch, which is manifested by cell cycle elongation, loss of synchronized divisions and genome activation. Moreover, depletion of Chmp4bb which is an essential ESCRT-III component for scission, delayed abscission beyond the MBT switch. Hallmark features of MBT, including transcription onset and cell shape changes, were similar in sibling cells connected by intercellular bridges, proposing a role for intercellular bridges in maintaining cell-cell communication in the embryo. Taken together, our data suggest that abscission is part of the cellular changes that occur during MBT and that cells coordinate their behavior during this critical embryonic phase through persisted intercellular bridges. Significance StatementIn this work we show that the last step of cytokinesis, termed abscission, is inhibited in early zebrafish embryos. As a result, sibling cells remain connected to one another for several cycles and mutually time their developmental progress including transcription onset. Abscission commences at the 10th cell cycle, when embryos enter the midblastula transition (MBT) switch in which embryonic cells become individualized and exhibit the characteristics of mature cells. Our data suggest that abscission is part of the MBT switch and that embryonic sibling cells mutually time their developmental progress by maintaining physical connections between them in the early embryo.

developmental biology↗

bioRxiv2020

Interpretable deep learning of label-free live cell images uncovers functional hallmarks of highly-metastatic melanoma

Deep convolutional neural networks have emerged as a powerful technique to identify hidden patterns in complex cell imaging data. However, these machine learning techniques are often criticized as uninterpretable "black-boxes" - lacking the ability to provide meaningful explanations for the cell properties that drive the machines prediction. Here, we demonstrate that the latent features extracted from label-free live cell images by an adversarial auto-encoding deep convolutional neural network capture subtle details of cell appearance that allow classification of melanoma cell states, including the metastatic efficiency of seven patient-derived xenograft models that reflect clinical outcome. Although trained exclusively on patient-derived xenograft models, the same classifier also predicted the metastatic efficiency of immortalized melanoma cell lines suggesting that the latent features capture properties that are specifically associated with the metastatic potential of a melanoma cell regardless of its origin. We used the autoencoder to generate "in-silico" cell images that amplified the cellular features driving the classifier of metastatic efficiency. These images unveiled pseudopodial extensions and increased light scattering as functional hallmarks of metastatic cells. We validated this interpretation by analyzing experimental image time-lapse sequences in which melanoma cells spontaneously transitioned between states indicative of low and high metastatic efficiency. Together, this data is an example of how the application of Artificial Intelligence supports the identification of processes that are essential for the execution of complex integrated cell functions but are too subtle to be identified by a human expert.

cell biology↗