Biology subjects

Segal, E.

Publications and source records attributed to Segal, E..

At least 19 recordsLinked to original sources

bioRxiv2026

Population-scale dietary signatures of the oral microbiome

Diet is an environmental factor that influences oral microbial communities. How specific foods, nutrients and dietary patterns relate to microbial diversity, species composition and functional potential remains incompletely characterised. Here, we combined detailed dietary records with buccal metagenomic profiles from 5,824 Human Phenotype Project participants. Using these data, we examined diet-oral microbiome associations with prediction models and covariate-adjusted analyses. Plant-based and guideline-based dietary patterns showed similar associations across oral microbial species. However, species within the same genus could show opposite associations with a dietary pattern. For example, a whole-food plant-based diet score was positively associated with Streptococcus sanguinis but negatively with S. gordonii and S. oralis. Some diet-oral microbiome associations were replicated at the genus level in an independent US National Health and Nutrition Examination Survey cohort of 4,667 participants. We map dietary associations with oral microbial diversity, species composition and functional potential, revealing shared microbial signatures across dietary patterns.

microbiology↗

bioRxiv2026

Image-Conditioned Diffusion for Privacy-Preserving Synthetic Medical Images

Medical imaging models depend on large, shareable datasets, yet privacy constraints limit data dissemination. Current text-conditioned diffusion models fail to preserve subtle, distributed clinical signals, such as continuous physiological biomarkers, rendering synthetic data insufficient for robust downstream physiological modeling. Here, we evaluate image-to-image (I2I) diffusion as a tunable, privacy-preserving transformation that produces a synthetic counterpart of real images while preserving downstream-relevant information. We fine-tune Stable Diffusion with low-rank adapters on retinal fundus photographs and chest radiographs, assessing fidelity, clinical signal preservation, cross-site transfer, and empirical re-identification risk. I2I consistently outperforms text-to-image generation in image fidelity and in preserving biomarker information. In cross-cohort transfer to an external retinal dataset from the UK Biobank, pretraining on I2I synthetic data performs comparably to real-image pretraining and surpasses it in the smallest fine-tuning sets. Varying I2I strength reveals that the privacy-utility tradeoff is highly modality-dependent: while retinal images achieve practical de-identification, chest X-rays exhibit structural combinatorics that leave them substantially re-identifiable even at high noise strengths, exposing critical boundaries for diffusion-based anonymization. These results position image-conditioned diffusion as a practical approach for generating shareable medical images with tunable de-identification.

bioinformatics↗

bioRxiv2025

Bacterial strains in the human gut associate with host physiology

The human microbiota influences host physiology, yet much of its complexity lies beneath the species level. Here, we analyzed the intra-species genetic diversity of 936 gut bacteria across 24,997 individuals from three countries. Our findings show that highly abundant species exhibit greater strain stability, whereas low-abundance species display increased in-clonal mutations. Clonal strains are often mutually exclusive, while genetically variable strains tend to coexist. Strain turnover is associated with the presence of annotated chemotaxis and sporulation genes in reference genomes, whereas strain coexistence is associated with quorum sensing and secretion systems. Leveraging deep phenotypic data, we constructed an atlas detailing strain-level associations with diverse host physiological domains. For example, Phocaeicola vulgatus sub-types relate to host obesity, Lachnospira eligens to sleep, and Parabacteroides distasonis to iron hemostasis. This resource may guide personalized microbiome-based interventions to improve human health.

microbiology↗

bioRxiv2025

Decoding the Oral Microbiome: Metagenomic Insights into Host Metabolic Health

The oral microbiome may capture system-specific information about host metabolic health, yet large-scale, multi-system evidence remains scarce. We analyzed 9,431 participants in the Human Phenotype Project (HPP), integrating buccal-swab oral whole-metagenome profiles with 44 metabolic measures spanning liver ultrasound, continuous glucose monitoring (CGM), and dual-energy X-ray absorptiometry (DXA). Using a microbiome-wide association study (MWAS) framework, we constructed a multilayer map across strains, gene families and pathways, revealing widespread associations: 213 strains, 124,603 gene families and 299 pathways were significantly associated with metabolic measures. Prioritizing the strongest and cross-phenotype signals, we identified multiple oral features with most significant associations to metabolic health. For example, acyl carrier protein (ACP) was associated with lower liver inflammation and reduced adiposity, whereas polyamine biosynthesis and ceramide -oxidation tracked higher glucose variability and adverse liver and adiposity phenotypes. Leveraging these MWAS-derived signals, we trained disease classification models using phenotype-selected oral features, which outperformed full-feature oral models across six metabolic diseases. These association signals were also robust in oral-health sensitivity analyses in HPP, and key BMI and waist-circumference associations directionally replicated at the genus level in an independent cohort (n = 20,293). Together, these findings provide a population-scale oral-metabolic association map and highlight the potential of oral microbial markers as non-invasive tools for metabolic risk stratification.

microbiology↗

bioRxiv2025

Self-supervised learning enables robust microbiome predictions in data-limited and cross-cohort settings

The gut microbiome plays a crucial role in human health, but machine learning applications in this field face significant challenges, including limited labeled data availability, high dimensionality, and batch effects across different cohorts. To address these limitations, we developed representation learning models for gut microbiome metagenomic data, drawing inspiration from foundation models approaches based on self-supervised and transfer learning principles. By leveraging a large collection of 85,364 metagenomic samples, we implemented multiple self-supervised learning methods, including masked autoencoders with varying masking rates and adapted single-cell RNA sequencing models (scVI and scGPT), to generate embeddings from bacterial abundance profiles. These learned representations demonstrated significant advantages over raw bacterial abundances in two key scenarios: first, when training predictive models with very limited labeled data, improving prediction performance for age (r = 0.14 vs. 0.06), BMI (r = 0.16 vs. 0.11), visceral fat mass (r = 0.25 vs. 0.18), and drug usage classification (PR-AUC = 0.81 vs. 0.73); and second, when generalizing predictions across different cohorts, consistently outperforming models based on raw abundances in cross-dataset evaluation. Our approach provides a valuable framework for leveraging self-supervised representation learning to overcome the data limitations inherent in microbiome research, potentially enabling more robust and generalizable machine learning applications in this field.

systems biology↗

bioRxiv2025

RobPicker: A Meta Learning Framework for Robust Identification of Macromolecules in Cryo-Electron Tomograms

Accurate particle picking of macromolecules of all kinds, shapes, and sizes in cryogenic electron tomograms (cryo-ET) is critical for understanding the molecular architecture of biological systems in their native state. Current deep learning methods have shown potential in identifying macromolecules from tomograms, but they are vulnerable to issues of noise in the training dataset obtained through human or automatic labeling and imbalanced distribution of macromolecule species. To address these limitations, we developed RobPicker, a meta-learning framework that effectively mitigates these issues by automatically learning deep neural networks to correct label errors and give greater emphasis to underrepresented macromolecule species. In evaluations across diverse cryo-ET datasets with noisy labels and imbalanced species distributions, RobPicker substantially outperforms state-of-the-art methods, particularly in identifying small and rare macromolecules. The efficiency and robustness of RobPicker can also be used for rapid fine-tuning of the tilt-series alignment, leading to improved tomogram reconstruction and enabling high-resolution cellular structural biology analysis.

bioinformatics↗

bioRxiv2025

Comprehensive Survey of Gut Microbiome Associations with Health Conditions in the Human Phenotype Project

The human gut microbiome is increasingly implicated with diverse health conditions, highlighting the importance of developing comprehensive resources to systematically map these associations. In this study, we leveraged the Human Phenotype Project (HPP) 10K cohort, a large, deeply phenotyped population study cohort with extensive metagenomic profiling, to explore associations between the gut microbiome and 37 health indications. These include conditions with established or emerging links to the microbiome, such as obesity, diabetes, hyperlipidemia, metabolic syndrome, inflammation, liver disease, cardiovascular and kidney conditions, immune and allergic diseases, gastrointestinal symptoms, mental health, and sleep-related traits. Using a curated set of approximately 200 refined phenotypic features and analyzing 1,184 microbial species, we performed robust statistical association analyses. We observed significant enrichment of associations in 14 health indications, with most associations reflecting increased microbial abundance in favorable health states, thereby suggesting potential microbial targets for intervention. This work introduces a publicly available, high-resolution resource to facilitate future research and support the development of microbiome-informed health strategies.

genomics↗

bioRxiv2024

Notch signaling maintains a progenitor-like subclass of hepatocellular carcinoma

Hepatocellular carcinomas (HCCs) constitute one of the few cancer indications for which mortality rates continue to rise. While Notch signaling dictates a key progenitor lineage choice during development, its role in HCC has remained controversial. Using therapeutic antibodies targeting Notch ligands and receptors to screen over 40 patient-derived xenograft models, we here identify progenitor-like HCCs that crucially depend on a tumor-intrinsic JAG1-NOTCH2 signal. Inhibiting this signal induces tumor regressions by triggering progenitor-to-hepatocyte differentiation, the same cell fate-switch that Notch controls during development. Transcriptomic analysis places the responsive tumors within the well-characterized progenitor subclass, a poor prognostic group of highly proliferative tumors, providing a diagnostic method to enrich for Notch-dependent HCCs. Furthermore, single-cell RNA sequencing uncovers a heterogeneous population of tumor cells and reveals how Notch inhibition shifts cells from a mixed cholangiocyte-hepatocyte lineage to one resembling mature hepatocytes. Analyzing the underlying transcriptional programs brings molecular detail to this process by showing that Notch inhibition de-represses expression of CEBPA, which enables the activity of HNF4, a hepatocyte lineage factor that is otherwise quiescent. We thus describe a compelling and targetable dependency in a poor-prognosis class of HCCs.

cancer biology↗

bioRxiv2024

Genetic underpinnings of predicted changes in cardiovascular function using self supervised learning

BackgroundThe genetic underpinnings of cardiovascular disease remain elusive. Contrastive learning algorithms have recently shown cutting-edge performance in extracting representations from electrocardiogram (ECG) signals that characterize cross-temporal cardiovascular state. However, there is currently no connection between these representations and genetics. MethodsWe designed a new metric, denoted as Delta ECG, which measures temporal shifts in patients cardiovascular state, and inherently adjusts for inter-patient differences at baseline. We extracted this measure for 4,782 patients in the Human Phenotype Project using a novel self-supervised learning model, and quantified the associated genetic signals with Genome-Wide-Association Studies (GWAS). We predicted the expression of thousands of genes extracted from Peripheral Blood Mononuclear Cells (PBMCs). Downstream, we ran enrichment and overrepresentation analysis of genes we identified as significantly predicted from ECG. FindingsIn a Genome-Wide Association Study (GWAS) of Delta ECG, we identified five associations that achieved genome-wide significance. From baseline embeddings, our models significantly predict the expression of 57 genes in men and 9 in women. Enrichment analysis showed that these genes were predominantly associated with the electron transport chain and the same immune pathways as identified in our GWAS. ConclusionsWe validate a novel method integrating self-supervised learning in the medical domain and simple linear models in genetics. Our results indicate that the processes underlying temporal changes in cardiovascular health share a genetic basis with CVD, its major risk factors, and its known correlates. Moreover, our functional analysis confirms the importance of leukocytes, specifically eosinophils and mast cells with respect to cardiac structure and function.

systems biology↗

bioRxiv2024

Pangenomes of Human Gut Microbiota Uncover Links Between Genetic Diversity and Stress Response

The human gut microbiota has been the subject of many studies, establishing its central role in host health and disease. Interplay between mutations, horizontal gene transfer, natural selection, and genetic drift, lead to genetic diversity within these species, resulting in different phenotypes and effects on the host. Pangenome represents the collective gene pool of different strains of the same species. Here, we created pangenomes for 728 human gut prokaryotic species, holding four times more genes than the highest quality individual genome, showcasing the genetic diversity inherent in the human gut population. We find these species have a core set of about a thousand genes that defines them, distinct even between closely related species, and an accessory set of genes that are unique to the different strains. Furthermore, we show a spectrum of microbial behavior, while some species exhibit a saturated or "closed" pangenome, suggesting a limited set of genetic capabilities, others maintain an "open" pangenome, indicating elevated adaptability through genetic diversity. We discover that high strain variability is associated with the capacity of species to undergo sporulation, whereas low strain variability is associated with carrying genes that facilitate antibiotic resistances, suggesting different evolutionary strategies for survival taken by these microbes. We further map the landscape of antibiotic resistance genes across the human gut population, and find 237 cases of extreme resistance, predominantly of Enterobacteriaceae species, even to last resort antibiotics kept for cases where traditional treatments have failed. Lastly, we associate microbial strain level differences with human age and sex, exemplifying how the presence of specific genes in Akkermansia muciniphila and Phocaeicola vulgatus relate to host characteristics. Overall, our research provides a comprehensive overview of the evolution, genetic complexity and functional potential of the human gut microbiota, emphasizing its significant implications for human health and disease. The pangenomes and the antibiotic resistances map constitute a valuable resource for further scientific research and therapeutic advancements.

microbiology↗

bioRxiv2024

Topographical changes in extracellular matrix during skin fibrosis and recovery can be evaluated using automated image analysis algorithms

Skin fibrosis is characterized by fibroblast activation and intradermal fat loss, resulting in excess deposition and remodeling of dermal extracellular matrix (ECM). The topography of the dominant ECM proteins, such as collagens, can indicate skin stiffness and remains understudied in evaluating fibrotic skin. Here, we adapted two different unbiased image analysis algorithms to define collagen topography and alignment in a genetically inducible and reversible Wnt activation fibrosis model. We demonstrated that Wnt activated fibrotic skin has altered collagen fiber characteristics and a loss of collagen alignment, which were restored in the reversible model. This study highlights how unbiased algorithms can be used to analyze ECM topography, providing novel avenues to evaluate fibrotic skin onset, recovery, and treatment.

cell biology↗

bioRxiv2024

Quantification of beta cell carrying capacity in prediabetes

Prediabetes, a subclinical state of high glucose, carries a risk of transition to diabetes. One cause of prediabetes is insulin resistance, which impairs the ability of insulin to control blood glucose. However, many individuals with high insulin resistance retain normal glucose due to compensation by enhanced insulin secretion by beta cells. Individuals seem to differ in their maximum compensation level, termed beta cell carrying capacity, such that low carrying capacity is associated with a higher risk of prediabetes and diabetes. Carrying capacity has not been quantified using a mathematical model and cannot be estimated directly from measured glucose and insulin levels in patients, unlike insulin resistance and beta cell function which can be estimated using HOMA-IR and HOMA-B formula. Here we present a mathematical model of beta cell compensation and carrying capacity, and develop a new formula called HOMA-C to estimate it from glucose and insulin measurements. HOMA-C estimates the maximal potential beta cell function of an individual, rather than the current beta cell function. We test this approach using longitudinal cohorts of prediabetic people, finding 10-fold variation in carrying capacity. Low carrying capacity is associated with higher risk of transitioning to diabetes. We estimate the timescales of beta cell compensation and insulin resistance using large datasets, showing that, unlike previous mathematical models, the new model can explain the slow rise in glucose over decades. Our mathematical understanding of beta cell carrying capacity may help to assess the risk of prediabetes in each individual.

systems biology↗

bioRxiv2024

Competitive fungal commensalism mitigates candidiasis pathology

The mycobiota are a critical part of the gut microbiome, but host-fungal interactions and specific functional contributions of commensal fungi to host fitness remain incompletely understood. Here we report the identification of a new fungal commensal, Kazachstania heterogenica var. weizmannii, isolated from murine intestines. K. weizmannii exposure prevented Candida albicans colonization and significantly reduced the commensal C. albicans burden in colonized animals. Following immunosuppression of C. albicans colonized mice, competitive fungal commensalism thereby mitigated fatal candidiasis. Metagenome analysis revealed K. weizmannii presence among human commensals. Our results reveal competitive fungal commensalism within the intestinal microbiota, independent of bacteria and immune responses, that could bear potential therapeutic value for the management of C. albicans-mediated diseases.

immunology↗

bioRxiv2023

A nonenzymatic dependency on inositol-requiring enzyme 1 controls cancer cell cycle progression and tumor growth

Endoplasmic-reticulum resident inositol-requiring enzyme 1 (IRE1) supports protein homeostasis via a cytoplasmic kinase-RNase module. Known cancer dependency on IRE1 entails its enzymatic activation of the transcription factor XBP1s and of RNA decay. We discovered that some cancer cells require IRE1 but not its enzymatic activity. IRE1 knockdown, but not enzymatic inhibition or XBP1 disruption, increased DNA damage and chromosome instability while engaging the TP53 pathway and cyclin-dependent kinase inhibitors and attenuating cell cycle progression. IRE1 depletion downregulated factors involved in chromosome replication and segregation and in chromatin remodeling. Immunoelectron microscopy indicated that endogenous IRE1 can localize to the nuclear envelope. Thus, cancer cells can require IRE1 either enzymatically or nonenzymatically, with significant implications for IRE1s biological role and therapeutic targeting.

cancer biology↗

bioRxiv2021

In-depth characterization of the serum antibody epitope repertoire in inflammatory bowel disease using phage-displayed immunoprecipitation sequencing

Inflammatory bowel diseases (IBD), e.g. Crohns disease (CD) and ulcerative colitis (UC), are chronic immune-mediated inflammatory diseases. A comprehensive overview of an IBD-specific antibody epitope repertoire is, however, lacking. We leveraged a high-throughput phage-displayed immunoprecipitation sequencing (PhIP-seq) workflow to identify antibodies against 344,000 antimicrobial, immune and food antigens in 497 IBD patients as compared to 1,326 controls. IBD was characterized by 373 differentially abundant antibodies (202 overrepresented and 171 underrepresented), with 17% shared by both IBDs, 55% unique to CD and 28% unique to UC. Antibodies against bacterial flagellins dominated in CD and were associated with ileal involvement, fibrostenotic disease and anti-Saccharomyces cerevisiae antibody positivity, but not with fecal microbiome composition. Antibody epitope repertoires accurately discriminated CD from controls (AUC=0.89), and similar discrimination was achieved when using only ten antibodies (AUC=0.87). IBD patients thus show a distinct antibody repertoire against selected peptides, allowing patient stratification and discovery of immunological targets.

immunology↗

bioRxiv2021

Genetic, environmental and intrinsic determinants of the human antibody epitope repertoire

Phage-displayed immunoprecipitation sequencing (PhIP-Seq) has successfully enabled high-throughput profiling of human antibody profiles. However, a comprehensive overview of environmental and genetic determinants shaping human adaptive immunity is currently lacking. In this study, we aimed to investigate the effects of genetic, environmental and intrinsic factors on the variation in human antibody repertoires. We characterized serological antibody repertoires against 344,000 peptides using PhIP-Seq libraries from a wide range of microbial and environmental antigens in 1,443 participants from a population cohort. We demonstrate individual-specificity, temporal consistency and co-housing similarities in antibody repertoire. Genetic analyses showed involvement of the HLA, IGHV and FUT2 regions. Furthermore, we uncovered associations between 48 phenotypic factors and 544 antibody-bound peptides, including age, cell counts, sex, smoking behavior and allergies, among others. Overall, our results indicate that human antibody epitope repertoires are shaped by both host genetics and environmental exposures and highlight unique signatures of distinct phenotypes and genotypes.

immunology↗

bioRxiv2021

Non-Coding Genetic Analysis Implicates Interleukin 18 Receptor Accessory Protein 3'UTR in Amyotrophic Lateral Sclerosis

The non-coding genome is substantially larger than the protein-coding genome but is largely unexplored by genetic association studies. Here, we performed region-based burden analysis of >25,000 variants in untranslated regions of 6,139 amyotrophic lateral sclerosis (ALS) whole-genomes and 70,403 non-ALS controls. We identified Interleukin-18 Receptor Accessory Protein (IL18RAP) 3'UTR variants significantly enriched in non-ALS genomes, replicated in an independent cohort, and associated with a five-fold reduced risk of developing ALS. Variant IL18RAP 3'UTR reduces mRNA stability and the binding of RNA-binding proteins. Variant IL18RAP 3'UTR further dampens neurotoxicity of human iPSC-derived C9orf72-ALS microglia that depends on NF-{kappa}B signaling. Therefore, the variant IL18RAP 3'UTR provides survival advantage for motor neurons co-cultured with C9-ALS microglia. The study reveals direct genetic evidence and therapeutic targets for neuro-inflammation, and emphasizes the importance of non-coding genetic association studies. One Sentence SummaryNon-coding genetic variants in IL-18 receptor 3UTR decrease ALS risk by modifying IL-18-NF-{kappa}B signaling in microglia.

neuroscience↗

bioRxiv2020

An atlas of robust microbiome associations with phenotypic traits based on large-scale cohorts from two continents

Numerous human conditions are associated with the microbiome, yet studies are inconsistent as to the magnitude of the associations and the bacteria involved, likely reflecting insufficiently employed sample sizes. Here, we collected diverse phenotypes and gut microbiota from 34,057 individuals from Israel and the U.S.. Analyzing these data using a much-expanded microbial genomes set, we derive an atlas of robust and numerous unreported associations between bacteria and physiological human traits, which we show to replicate in cohorts from both continents. Using machine learning models trained on microbiome data, we predict human traits with high accuracy across continents. Subsampling our cohort to smaller cohort sizes yielded highly variable models and thus sensitivity to the selected cohort, underscoring the utility of large cohorts and possibly explaining the source of discrepancies across studies. Finally, many of our prediction models saturate at these numbers of individuals, suggesting that similar analyses on larger cohorts may not further improve these predictions.

genomics↗