Biology subjects

Mahmood, F.

Publications and source records attributed to Mahmood, F..

13 recordsLinked to original sources

bioRxiv2026

Prototype-based AI triage for 3D pathology

Non-destructive 3D pathology enables high-resolution slide-free imaging of intact clinical specimens, providing comprehensive visualization of tissue structures beyond what conventional slide-based 2D histopathology can provide. However, the scale and complexity of volumetric datasets make exhaustive manual review impractical, motivating AI-assisted triage methods to select a small number of high-risk 2D slices for pathologist review. While prior triage models have shown promise, interpretability is poor and performance can be suboptimal, especially in the nascent field of 3D pathology in which labeled data is limited. We present SCOPE, a Segmentation-guided CrOss-slice PrototypE learning framework for comprehensive risk assessment of 2D levels within 3D pathology datasets. SCOPE combines (i) clustering-based pretraining on large-scale unlabeled volumetric data to initialize morphology-aware prototypes, (ii) segmentation-derived structural priors from publicly available models to guide proto-type learning, and (iii) cross-slice (2.5D) prototype aggregation across neighboring slices to generate slice-level risk predictions. In prostate and esophageal data cohorts, SCOPE consistently outperforms attention-based and prototype-based multiple instance learning baselines for both binary and multiclass prediction tasks, enabling depth-resolved risk profiling for 3D triage based on morphological prototypes that are interpretable to pathologists.

pathology↗

bioRxiv2026

Scalable 3D cell-interaction analysis via supercell graphs for prostate cancer risk stratification

Cellular interactions underlie fundamental biological processes but are not fully represented in conventional 2D histology images. While 3D pathology allows for more-accurate construction of cell-level graphs, machine-learning models are computationally unwieldy and prone to overfitting, especially when dealing with small cohorts. Here, we introduce SCALE3D, a SuperCell graph Analysis framework for LargE 3D pathology datasets. In SCALE3D, spatially adjacent and morphologically similar cells are grouped into functional "supercells." Supercell subtypes are defined via morphology-based clustering and 3D graphs connecting these supercells are used to model their interactions. Validation was performed with 76 radical prostatectomy specimens from patients with known 5-year biochemical recurrence (BCR) outcomes. SCALE3D-derived features achieve higher performance for BCR prediction than established 3D nuclear and glandular morphological features. Combining these complementary features further improves prediction performance. Compared to individual cell-level 3D graphs, SCALE3D maintains comparable prognostic performance with improved noise tolerance while reducing computational times by up to 1,000-fold.

pathology↗

bioRxiv2025

Pesticide-exposed bees fail to thermoregulate leading to cold colonies with consequences for offspring development

For many organisms, effective thermoregulation is needed to cope with changing environmental temperatures. But if exposure to pesticides in the environment were to impair this homeostatic process, reproduction and population viability could be at risk. Focusing on an important insect pollinator, we conducted three complementary experiments exposing bumblebees to a pesticide under different temperature challenges and measuring the impacts on thermoregulatory ability and brood development. First, we reveal that pesticide-exposed individuals cannot maintain a stable thorax temperature, especially at lower temperatures. Second, reductions in body temperature are accompanied by behavioural changes and that pesticide-exposed colonies fail to maintain appropriate brood temperatures. Third, such collective impairment on brood thermoregulation (not the pesticide toxicity to offspring per se) leads to delayed pupal development and reduced adult population growth. Our study provides a valid mechanistic explanation for why terrestrial insects requiring brood thermoregulation have declined. With frequent extreme weather events forecasted, our findings have concerning implications for how populations will adequately persist and grow under current pesticide-use regimes with ramifications on pollination services.

ecology↗

bioRxiv2025

6S RNA facilitates bacterial virulence and adaptation at the epithelial barrier

Enteropathogenic bacteria must adapt dynamically to the complex gastrointestinal environment to successfully colonize host tissue and evade immune defenses. Using Yersinia pseudotuberculosis as a model, we performed in vivo spatial transcriptomics to investigate bacterial gene expression as it translocates from cecal lumen to associated lymphoid tissue in the mouse intestine. By optimizing bacterial RNA enrichment, we achieved near-complete transcriptome coverage and identified compartment-specific transcriptional profiles. Bacteria in lymphoid tissue exhibited elevated expression of virulence-associated type III secretion system (T3SS) genes and markers of increased replication, alongside a higher plasmid copy number. Oxygen availability emerged as an environmental cue for T3SS induction. Importantly, we discovered a role for the non-coding 6S RNA in accelerating virulence gene expression. Tissue-localized bacteria had significantly upregulated 6S RNA levels, and deletion of the ssrS gene encoding 6S RNA impaired T3SS gene expression and effector secretion. Further, spatial analyses of bacterial gene expression in foci of infected lymphoid tissue revealed heterogeneous expression patterns with significantly elevated expression of T3SS and 6S RNA in bacteria located close to surrounding phagocytes. Together these findings demonstrate that Y. pseudotuberculosis undergoes rapid transcriptional reprogramming upon epithelial barrier crossing coordinated by environmental sensing, where 6S RNA accelerates tissue colonization by promoting efficient expression of T3SS.

microbiology↗

bioRxiv2025

Whole-organism screening in a zebrafish model of CLN2 disease identifies pregnenolone as a modulator of lysosomal functions with anti-epileptic properties

Lysosomal storage disorders (LSDs), a group of inherited genetic diseases, are often associated with early-onset neurodegeneration and refractory epileptic seizures. In CLN2 disease, an LSD caused by recessively inherited dysfunction of lysosomal serine protease Tripeptidyl Peptidase 1 (TPP1), lysosomes are functionally impaired through a characteristic accumulation of subcellular materials. Here, we develop and apply a whole-organism screening workflow in tpp1-/- zebrafish to identify small molecules that suppress epileptic seizures - a hallmark of the human disease - in this model. Among 640 US Food and Drug Administration-approved drugs, pregnenolone, an endogenous precursor for steroid biosynthesis, efficiently suppresses seizures and cell death in tpp1-/- zebrafish. Using a semi-automated high-content workflow, we further show that pregnenolone normalizes lysosomal architecture in tpp1-/- zebrafish. Pregnenolone stimulates steroid hormone biosynthesis and related gene expression, which is dysregulated in tpp1-/- zebrafish. Taken together, tpp1-/- zebrafish are a suitable model to study CLN2 disease, in which we have identified pregnenolone as a candidate with therapeutic properties.

pharmacology and toxicology↗

bioRxiv2025

Deep-learning triage of 3D pathology datasets for comprehensive and efficient pathologist assessments

Standard-of-care slide-based 2D histopathology severely undersamples spatially heterogeneous tissue specimens, with each thin 2D section representing <1% of the entire tissue volume (in the case of a biopsy). Recent advances in non-destructive 3D pathology, such as open-top light-sheet microscopy (OTLS), enable comprehensive high-resolution imaging of large clinical specimens. While fully automated computational analyses of such 3D pathology datasets are being explored, a potential low-risk route for accelerated clinical adoption would be to continue to rely upon pathologists to provide final diagnoses. Since manual review of these massive and complex 3D datasets is infeasible for routine clinical practice, we present CARP3D, a deep learning triage framework that identifies high-risk 2D cross sections within large 3D pathology datasets to enable time-efficient pathologist evaluation. CARP3D assigns risk scores to all 2D levels within a tissue volume by leveraging context from a subset of neighboring depth levels, outperforming models in which predictions are based on isolated 2D levels. In two use cases - risk stratification based on prostate cancer biopsies and screening for dysplasia/cancer in endoscopic biopsies of Barretts esophagus - AI-triaged 3D pathology, enabled by CARP3D, demonstrates the potential to improve the detection of high-risk diseases in comparison to slide-based 2D histopathology while optimizing pathologist workloads.

pathology↗

bioRxiv2024

AI-driven Discovery of Morphomolecular Signatures in Toxicology

Early identification of drug toxicity is essential yet challenging in drug development. At the preclinical stage, toxicity is assessed with histopathological examination of tissue sections from animal models to detect morphological lesions. To complement this analysis, toxicogenomics is increasingly employed to understand the mechanism of action of the compound and ultimately identify lesion-specific safety biomarkers for which in vitro assays can be designed. However, existing works that aim to identify morphological correlates of expression changes rely on qualitative or semi-quantitative morphological characterization and remain limited in scale or morphological diversity. Artificial intelligence (AI) offers a promising approach for quantitatively modeling this relationship at an unprecedented scale. Here, we introduce GEESE, an AI model designed to impute morphomolecular signatures in toxicology data. Our model was trained to predict 1,536 gene targets on a cohort of 8,231 hematoxylin and eosin-stained liver sections from Rattus norvegicus across 127 preclinical toxicity studies. The model, evaluated on 2,002 tissue sections from 29 held-out studies, can yield pseudo-spatially resolved gene expression maps, which we correlate with six key drug-induced liver injuries (DILI). From the resulting 25 million lesion-expression pairs, we established quantitative relations between up and downregulated genes and lesions. Validation of these signatures against toxicogenomic databases, pathway enrichment analyses, and human hepatocyte cell lines asserted their relevance. Overall, our study introduces new methods for characterizing toxicity at an unprecedented scale and granularity, paving the way for AI-driven discovery of toxicity biomarkers. Live demo: https://mahmoodlab.github.io/tox-discovery-ui/

bioinformatics↗

bioRxiv2024

Deep Learning-based Modeling for Preclinical Drug Safety Assessment

In drug development, assessing the toxicity of candidate compounds is crucial for successfully transitioning from preclinical research to early-stage clinical trials. Drug safety is typically assessed using animal models with a manual histopathological examination of tissue sections to characterize the dose-response relationship of the compound - a timeintensive process prone to inter-observer variability and predominantly involving tedious review of cases without abnormalities. Artificial intelligence (AI) methods in pathology hold promise to accelerate this assessment and enhance reproducibility and objectivity. Here, we introduce TRACE, a model designed for toxicologic liver histopathology assessment capable of tackling a range of diagnostic tasks across multiple scales, including situations where labeled data is limited. TRACE was trained on 15 million histopathology images extracted from 46,734 digitized tissue sections from 157 preclinical studies conducted on Rattus norvegicus. We show that TRACE can perform various downstream toxicology tasks spanning histopathological response assessment, lesion severity scoring, morphological retrieval, and automatic dose-response characterization. In an independent reader study, TRACE was evaluated alongside ten board-certified veterinary pathologists and achieved higher concordance with the consensus opinion than the average of the pathologists. Our study represents a substantial leap over existing computational models in toxicology by offering the first framework for accelerating and automating toxicological pathology assessment, promoting significant progress with faster, more consistent, and reliable diagnostic processes. Live Demo: https://mahmoodlab.github.io/tox-foundation-ui/

bioinformatics↗

bioRxiv2024

Epstein-Barr Virus Orchestrates Spatial Reorganization and Immunomodulation within the Classic Hodgkin Lymphoma Tumor Microenvironment

Classic Hodgkin Lymphoma (cHL) is a tumor composed of rare malignant Hodgkin and Reed-Sternberg (HRS) cells nested within a T-cell rich inflammatory immune infiltrate. cHL is associated with Epstein-Barr Virus (EBV) in 25% of cases. The specific contributions of EBV to the pathogenesis of cHL remain largely unknown, in part due to technical barriers in dissecting the tumor microenvironment (TME) in high detail. Herein, we applied multiplexed ion beam imaging (MIBI) spatial pro-teomics on 6 EBV-positive and 14 EBV-negative cHL samples. We identify key TME features that distinguish between EBV-positive and EBV-negative cHL, including the relative predominance of memory CD8 T cells and increased T-cell dysfunction as a function of spatial proximity to HRS cells. Building upon a larger multi-institutional cohort of 22 EBV-positive and 24 EBV-negative cHL samples, we orthogonally validated our findings through a spatial multi-omics approach, coupling whole transcriptome capture with antibody-defined cell types for tu-mor and T-cell populations within the cHL TME. We delineate contrasting transcriptomic immunological signatures between EBV-positive and EBV-negative cases that differently impact HRS cell proliferation, tumor-immune interactions, and mecha-nisms of T-cell dysregulation and dysfunction. Our multi-modal framework enabled a comprehensive dissection of EBV-linked reorganization and immune evasion within the cHL TME, and highlighted the need to elucidate the cellular and molecular fac-tors of virus-associated tumors, with potential for targeted therapeutic strategies.

cancer biology↗

bioRxiv2023

Deep Learning Model Imputes Missing Stains in Multiplex Images

Multiplex staining enables simultaneous detection of multiple protein markers within a tissue sample. However, the increased marker count increased the likelihood of staining and imaging failure, leading to higher resource usage in multiplex staining and imaging. We address this by proposing a deep learning-based MArker imputation model for multipleX IMages (MAXIM) that accurately impute protein markers by leveraging latent biological relationships between markers. The models imputation ability is extensively evaluated at pixel and cell levels across various cancer types. Additionally, we present a comparison between imputed and actual marker images within the context of a downstream cell classification task. The MAXIM models interpretability is enhanced by gaining insights into the contribution of individual markers in the imputation process. In practice, MAXIM can reduce the cost and time of multiplex staining and image acquisition by accurately imputing protein markers affected by staining issues.

pathology↗

bioRxiv2023

MAPS: Pathologist-level cell type annotation from tissue images through machine learning

Highly multiplexed protein imaging is emerging as a potent technique for analyzing protein distribution within cells and tissues in their native context. However, existing cell annotation methods utilizing high-plex spatial proteomics data are resource intensive and necessitate iterative expert input, thereby constraining their scalability and practicality for extensive datasets. We introduce MAPS (Machine learning for Analysis of Proteomics in Spatial biology), a machine learning approach facilitating rapid and precise cell type identification with human-level accuracy from spatial proteomics data. Validated on multiple in-house and publicly available MIBI and CODEX datasets, MAPS outperforms current annotation techniques in terms of speed and accuracy, achieving pathologist-level precision even for challenging cell types, including tumor cells of immune origin. By democratizing rapidly deployable and scalable machine learning annotation, MAPS holds significant potential to expedite advances in tissue biology and disease comprehension.

cancer biology↗

bioRxiv2022

Chemical engineering of therapeutic siRNAs for allele-specific gene silencing in vivo in CNS

Small interfering RNAs (siRNAs) are a new class of drugs, exhibiting sequence-driven, potent, and sustained silencing of gene expression in vivo. We recently demonstrated that siRNA chemical architectures can be optimized to provide efficient delivery to the CNS. Many genetically-defined neurodegenerative disorders are autosomal dominant favoring selective silencing of the mutant allele. In some cases, successful targeting of the mutant allele requires targeting of a single nucleotide polymorphism (SNP) heterozygosity. Using Huntingtons disease as a model, we demonstrate allele-specific RNAi-based silencing of gene expression in vivo and in neurons differentiated from HD patient-derived iPSCs. A series of in vitro screens, with chemical and thermodynamic optimization, identified compounds with >50-fold selectivity for the mutant HD-causing allele, based on a single nucleotide difference. The optimized compound exhibits selective silencing of mutant huntingtin (HTT) protein in patient derived cells and throughout the HD mouse brain, providing a demonstration of SNP-based allele-specific RNAi silencing of gene expression in vivo in the CNS. The ability to target a disease-causing allele using RNAi-based therapies could be applied to a wide range of dominant CNS disorders, where maintenance of wild-type expression is essential.

neuroscience↗

bioRxiv2021

Benchmarking artificial intelligence methods for end-to-end computational pathology

Artificial intelligence (AI) can extract subtle visual information from digitized histopathology slides and yield scientific insight on genotype-phenotype interactions as well as clinically actionable recommendations. Classical weakly supervised pipelines use an end-to-end approach with residual neural networks (ResNets), modern convolutional neural networks such as EfficientNet, or non-convolutional architectures such as vision transformers (ViT). In addition, multiple-instance learning (MIL) and clustering-constrained attention MIL (CLAM) are being used for pathology image analysis. However, it is unclear how these different approaches perform relative to each other. Here, we implement and systematically compare all five methods in six clinically relevant end-to-end prediction tasks using data from N=4848 patients with rigorous external validation. We show that histological tumor subtyping of renal cell carcinoma is an easy task which approaches successfully solved with an area under the receiver operating curve (AUROC) of above 0.9 without any significant differences between approaches. In contrast, we report significant performance differences for mutation prediction in colorectal, gastric and bladder cancer. Weakly supervised ResNet-and ViT-based workflows significantly outperformed other methods, in particular MIL and CLAM for mutation prediction. As a reason for this higher performance we identify the ability of ResNet and ViT to assign high prediction scores to highly informative image regions with plausible histopathological image features. We make all source codes publicly available at https://github.com/KatherLab/HIA, allowing easy application of all methods on any end-to-end problem in computational pathology.

bioinformatics↗