Biology subjects

Bahrami, M.

Publications and source records attributed to Bahrami, M..

14 recordsLinked to original sources

bioRxiv2026

Recruitment of the local GABAergic population engages complex network interactions in the ventral subiculum

The ventral subiculum (vSub), a major output node of the hippocampal formation, has been implicated in diverse neurological disorders. Disruption of excitatory-inhibitory balance is a common feature of neurological disorders, with altered GABAergic signaling representing a critical component. More recently, the vSub has gained increasing attention as a component of neural circuits underlying emotional and motivational regulation, through its extensive connections with regions implicated in mental health disorders such as addiction and anxiety. Our previous work in models of alcohol use disorder and comorbid anxiety identified an important role for GABAergic inhibition in regulating vSub excitability, prompting us to investigate the functional consequences of recruiting the broader local GABAergic population. Although previous studies have examined individual interneuron subtypes, the network consequences of activating the global population of local GABAergic neurons remain poorly understood. To address this question, we expressed channelrhodopsin using an mDlx-dependent strategy to target GABAergic neurons in the ventral hippocampus and performed in vivo electrophysiological recordings from vSub neurons in anesthetized rats. Repeated activation of this GABAergic population at 20 Hz, a stimulation frequency commonly used in behavioral optogenetic studies, produced robust suppression of pyramidal neuron firing during stimulation. This direct inhibition was accompanied by network-dependent increases in post-stimulus activity that varied as a function of baseline activity, sex, and trial number. Complementary ex vivo electrophysiological experiments further supported network-dependent effects and revealed short-term synaptic plasticity that evolved across repeated stimulation trials, reflecting contributions from both pre- and postsynaptic mechanisms. Together, these findings demonstrate that recruitment of the broader GABAergic population produces effects extending beyond direct inhibition and can engage dynamic network responses that depend on stimulation history and circuit state. These findings highlight the importance of stimulation frequency, timing, and network context when interpreting the functional consequences of GABAergic circuit manipulation.

neuroscience↗

bioRxiv2026

Benchmarking gene expression reconstruction from single-cell latent representations

Single-cell transcriptomics is typically modeled in low-dimensional latent representations that improve the signal-to-noise ratio of the data. Such representations underpin data integration, cell state discovery, and perturbation prediction, with applications ranging from large-scale organ atlases to latent trajectory modeling. Recent virtual cell approaches further leverage these representations to predict cellular responses as distributional shifts in latent space. Each of these applications ultimately requires faithful gene expression reconstruction from latent spaces for biological interpretation, enabling gene-level analysis of predicted perturbed or batch-corrected cells. Yet representation choice is typically treated as an implementation detail rather than a primary modeling decision, with no systematic evaluation of how well latent representations support gene expression reconstruction. Here, we introduce ReconEval, a benchmark for evaluating gene expression reconstruction from single-cell latent spaces. We benchmark two classes of latent representations: end-to-end trained models such as PCA, autoencoders, and variational autoencoders, and pretrained single-cell foundation model embeddings coupled to newly trained decoders. Reconstruction is evaluated both directly and after latent-space perturbation prediction. Across perturbational and observational datasets totaling over 100 million cells, our metric suite quantifies statistical fidelity; biological signal preservation, including differential expression, coexpression, cell-cycle structure, cytokine response and pathway activity; and perturbation-specific effects. We find that autoencoders achieve the strongest stand-alone reconstruction at low dimensionality, while variational regularization does not improve generalization in reconstruction. Frozen foundation model embeddings retain recoverable gene-level information, with reconstruction quality depending strongly on decoder architecture and pretraining objective. In latent perturbation modeling, high-dimensional PCA matches foundation model embeddings, while low-dimensional AE embeddings are optimal for flow-based generative models. Overall, reconstruction depends critically on the interplay between representation and downstream model, and simpler representations can outperform complex alternatives given appropriate capacity. Our benchmark establishes reconstruction as a critical evaluation axis for single-cell foundation models. We envision it improving the biological interpretability of latent-space modeling, a prerequisite for future virtual cell models to be validated by domain experts and grounded in biology.

bioinformatics↗

bioRxiv2026

Cellpin enables reference-based imputation and denoising of spatial transcriptomes

Spatially resolved transcriptomics enables gene expression profiling within tissue architecture, but targeted panels leave much of the transcriptome unmeasured and spatial artifacts such as RNA diffusion and segmentation errors introduce technical noise. These limitations necessitate computational imputation and denoising, yet existing methods typically incorporate spatial measurements during training, limiting scalability and risking the embedding of technology-specific artifacts into learned representations. To address this, we present cellpin, a variational autoencoder trained exclusively on single-cell RNA sequencing data, using teacher-student latent distillation and noise-simulating augmentations to jointly impute unmeasured genes and denoise spatial profiles without requiring cross-modality alignment. Benchmarked against six methods across multiple paired datasets, cellpin achieves superior held-out gene prediction while scaling efficiently to atlas-size references and multi-sample cohorts. In full-transcriptome Atera data, cellpin reduces residual spatial noise and improves cell-state resolution, providing a scalable and principled foundation for biological discovery from spatial transcriptomics data.

bioinformatics↗

bioRxiv2026

The individuality of single-frame functional brain connectivity

AbstractConverging evidence from studies on brain network "fingerprinting" and precision functional mapping suggest that brain networks are highly individualized in functionally meaningful ways. Concurrently with a growth in studies on this topic, there has been a rise in interest on dynamics (approximately second-to-second changes) in brain networks within scan sessions. While analyses of traditional static networks have increasingly grown towards emphasizing the importance of individual differences in brain network topology, studies of dynamic networks typically follow methodology that require brain states to be considered at a group level. Recent studies have begun to assess the individuality of recurring dynamic brain "states". In this work, we extend this recent work by exploring the extent to which functional connectivity fingerprinting is feasible at single-frame temporal resolution. We estimate connectivity at individual volumes using phase coherence. We find that the identity of participants can be classified based on single volumes given sufficient database scan data and that having more highly parcellated atlases facilitates identification. Finally, we find that tasks can be identified more readily within subjects than between subjects. We conclude that participant identity may be an important driver of observed single-volume connectivity patterns. Further, the single-volume neural correlates of a task appear to be more consistent within subjects than between subjects. This highlights the importance of considering individual variability in studies of brain network dynamics.

neuroscience↗

bioRxiv2025

scConcept: Contrastive pretraining for technology-agnostic single-cell representations beyond reconstruction

Recent large-scale single-cell foundation models have shown promise for exploring cellular states, yet they often underperform compared to simpler, domain-specific methods, raising concerns about their broader applicability. A key limitation lies in their reliance on masked language modeling, which is well suited for generative language tasks but poorly aligned with learning rich cell-level embeddings required in single-cell biology. Moreover, the proliferation of transcriptomic technologies--from whole transcriptome dissociated assays to image-based targeted profiling--poses a major challenge for cross-platform generalization. Here, we align with recent advances in machine learning to move beyond reconstruction metrics, which often do not capture important sample variation. We present scConcept ("contrastive cell pre-training"), a transformer-based contrastive learning framework that directly optimizes cell embeddings by contrasting multiple views of cells. By replacing gene-level reconstruction with a cell-level identification task, scConcept learns robust representations that are invariant to count distributions and gene panel selection, across diverse assays and technologies. To highlight the capability of the proposed framework, we pretrain scConcept on a similar corpus of over 30 million single-cell RNA-seq profiles as recent foundation models. Our approach demonstrates superior performance not only compared to state-of-the-art pretrained foundation models but also domain-specific methods in various downstream tasks, including cell-type annotation, technology integration, dissociated to spatial cell-type transfer, spatial imputation, gene panel optimization, and mapping new technologies on already existing atlases. Our results highlight contrastive pretraining as a powerful alternative to reconstruction-based strategies for single-cell modeling, providing a path toward general-purpose, technology-agnostic cell representations.

bioinformatics↗

bioRxiv2024

Pertpy: an end-to-end framework for perturbation analysis

Advances in single-cell technology have enabled the measurement of cell-resolved molecular states across a variety of cell lines and tissues under a plethora of genetic, chemical, environmental, or disease perturbations. Current methods focus on differential comparison or are specific to a particular task in a multi-condition setting with purely statistical perspectives. The quickly growing number, size, and complexity of such studies requires a scalable analysis framework that takes existing biological context into account. Here, we present pertpy, a Python-based modular framework for the analysis of large-scale perturbation single-cell experiments. Pertpy provides access to harmonized perturbation datasets and metadata databases along with numerous fast and user-friendly implementations of both established and novel methods such as automatic metadata annotation or perturbation distances to efficiently analyze perturbation data. As part of the scverse ecosystem, pertpy interoperates with existing libraries for the analysis of single-cell data and is designed to be easily extended.

bioinformatics↗

bioRxiv2024

Contrasting topologies of synchronous and asynchronous functional brain networks

We generated asynchronous functional networks (aFNs) using a novel method called optimal causation entropy (oCSE) and compared aFN topology to the correlation-based synchronous functional networks (sFNs) which are commonly used in network neuroscience studies. Functional magnetic resonance imaging (fMRI) time series from 212 participants of the National Consortium on Alcohol and NeuroDevelopment in Adolescence (NCANDA) study were used to generate aFNs and sFNs. As a demonstration of how aFNs and sFNs can be used in tandem, we used multivariate mixed effects models to determine whether age interacted with node efficiency to influence connection probabilities in the two networks. After adjusting for differences in network density, aFNs had higher global efficiency but lower local efficiency than the sFNs. In the aFNs, nodes with the highest outgoing global efficiency tended to be in the brainstem and orbitofrontal cortex. aFN nodes with the highest incoming global efficiency tended to be members of the Default Mode Network (DMN) in sFNs. Age interacted with node global efficiency in aFNs and node local efficiency in sFNs to influence connection probability. We conclude that the sFN and aFN both offer information about functional brain connectivity which the other type of network does not.

neuroscience↗

bioRxiv2024

Nicheformer: a foundation model for single-cell and spatial omics

Tissue makeup relies fundamentally on the cellular microenvironment. Spatial single-cell genomics allows probing the underlying cellular interactions in an unbiased, scalable fashion. To learn a unified cell representation that accounts for local dependencies in the cellular microenvironment, we propose Nicheformer, a transformer-based foundation model that combines human and mouse dissociated single-cell and targeted spatial transcriptomics data. Pretrained on over 57 million dissociated and 53 million spatially resolved cells across 73 tissues on cellular reconstruction, the model is fine-tuned on spatial tasks for spatial omics data to decode spatially resolved cellular information. Nicheformer excels in linear-probing and fine-tuning scenarios for a novel set of downstream tasks, in particular spatial composition prediction and spatial label prediction. We further show that existing foundation models trained on dissociated single-cell data alone are not capable of recapitulating the spatial complexity of cells in their microenvironments, indicating that multiscale models are required to understand complex local dependencies at scale. Nicheformer enables the prediction of the spatial context of dissociated cells, allowing the transfer of rich spatial information to scRNA-seq datasets. Overall, Nicheformer sets the stage for the next generation of machine-learning models in spatial single-cell analysis. Extended AbstractTissue makeup and the corresponding orchestration of vital biological activities, ranging from development and differentiation to immune response and regeneration, rely fundamentally on the cellular microenvironment and the interactions between cells. Spatial single-cell genomics allows probing such interactions in an unbiased and, increasingly, scalable fashion. To learn a unified cell representation that accounts for local dependencies in the cellular microenvironment and the underlying cell interactions, we propose to generalize recent foundation modeling approaches for disassociated single-cell transcriptomics to the spatial omics setting. Our model, Nicheformer, is a transformer-based foundation model that combines human and mouse dissociated single-cell and targeted spatial transcriptomics data to learn a cellular representation useful for a large variety of downstream tasks. Nicheformer is pretrained on over 57 million dissociated and 53 million spatially resolved cells across 73 tissues from both human and mouse. Subsequently, the model is fine-tuned on spatial tasks for spatial omics data to decode spatially resolved cellular information. We demonstrate the usefulness of Nicheformer in both linear-probing as well as fine-tuning scenarios on a novel set of spatially-relevant downstream tasks such as spatial density prediction or niche and region label prediction. In particular, we show that Nicheformer enables the prediction of the spatial context of dissociated cells, allowing the transfer of rich spatial information to scRNA-seq datasets. We define a series of novel spatial prediction problems and observe consistent top performance of Nicheformer, demonstrating the advantage of the improved model capacity of the underlying transformer. Additionally, we benchmarked Nicheformer in these tasks against scGPT1, Geneformer2, scVI3 and PCA and show that the Nicheformer architecture excels in these tasks. Altogether, our large-scale resource of more than 110 million cells in a partial spatial context, together with the set of novel spatial learning tasks and the Nicheformer model itself, will pave the way for the next generation of machine-learning models for spatial single-cell analysis.

bioinformatics↗

bioRxiv2024

CRISPR/Cas9-mediated deletion of a kinetoplast-associated gene attenuates virulence in Leishmania major parasites

The CRISPR/Cas9 system has emerged as a powerful tool for precise genome editing, allowing for the deletion of genes, generation of point mutations, and addition of tags to endogenous genes. We employed an efficient CRISPR/Cas9 technique in Leishmania major to assess its efficiency in editing a kDNA-associated gene, universal minicircle sequence binding protein (UMSBP), which is involved in mitochondrial respiration and kinetoplast division. We generated UMSBP C-tagged and UMSBP single knockout L. major (LmUMSBP+/-) parasites using the CRISPR/Cas9 toolkit. C-tagged parasite were confirmed by PCR, flow cytometry and Western blot analyses. Gene expression of mitochondrial redox regulating enzymes, tryparedoxin peroxidase (TXNPx) and trypanothione synthetase (TryS), were analysed by real-time RT-PCR. Growth rate of promastigotes in culture and infectivity rate in macrophages were analysed in vitro. Mice were immunized by LmUMSBP+/- mutant strain and lesion size and parasite burden were measured upon challenge with live wild type (WT) L. major. Cytokines were titrated on supernatant of lymph nodes cell culture by sandwich ELISA. Complete UMSBP deletion (LmUMSBP-/- null mutant) impaired promastigote survival, suggesting its essential role in parasite fitness. Despite this, we were able to produce attenuated LmUMSBP+/- parasites, which showed significant reduced growth in culture (P<0.05), increase in apoptosis (P<0.05) and downregulation of TXNPx and TryS gene expressions during growth of promastigotes compared to WT L. major. LmUMSBP+/- mutant strains did not cause ulcerative lesions in susceptible BALB/c mouse model. Furthermore, immunization of mice with LmUMSBP+/- parasites elicited a Th1 immune response with significantly high IFN-{gamma} and low IL-4 production in cell culture (P<0.001) associated with partial protection against L. major WT challenge, as evidenced by reduced parasite burden and lesion development in BALB/c mice. Our findings demonstrate the potential of CRISPR/Cas9-edited LmUMSBP+/- parasites as live attenuated vaccine candidate against Leishmania infection. Author summaryIn this study, we utilized the powerful CRISPR/Cas9 technique to edit the genome of Leishmania major, a parasite responsible for causing leishmaniasis. Specifically, we targeted a gene called universal minicircle sequence binding protein (UMSBP), which plays a crucial role in the parasites mitochondrial function and replication. Using CRISPR/Cas9, we successfully created two types of parasites: one with a tagged UMSBP gene and another with the UMSBP gene completely knocked out. We produced an attenuated parasites with deleting UMSBP gene having reduced growth and increased apoptosis compared to wild-type parasites. Importantly, immunizing mice with these attenuated parasites induced a strong immune response, particularly IFN-{gamma} secretion, and provided partial protection against infection with wild-type parasites. Our study suggests that CRISPR/Cas9-edited parasites could serve as promising live attenuated vaccine candidates against leishmaniasis.

immunology↗

bioRxiv2024

Delineating the Effective Use of Self-Supervised Learning in Single-Cell Genomics

Self-supervised learning (SSL) has emerged as a powerful method for extracting meaningful representations from vast, unlabeled datasets, already transforming computer vision and natural language processing. Similarly, in single-cell genomics (SCG), representation learning is well-recognized for offering insights into complex biological data, even more so by the advent of early foundation model approaches. However, despite these advancements, identifying scenarios in SCG where SSL outperforms traditional supervised or unsupervised learning methods remains a nuanced challenge. Furthermore, selecting the most effective pretext tasks within the SSL framework for SCG is a critical yet unresolved question. Here, we address this gap by adapting and benchmarking SSL techniques in SCG, including masked autoencoders with multiple masking strategies and contrastive learning approaches. Trained on over 20 million cells, this study rigorously examines multiple downstream tasks, including cell type prediction, gene expression reconstruction, cross-modality prediction, and data integration. Our empirical analyses underscore the nuanced role of SSL, namely in transfer learning scenarios leveraging auxiliary data or analyzing novel datasets. Masked autoencoders excel over contrastive methods in SCG, diverging from computer vision trends. Moreover, our findings reveal notable capabilities of SSL in zero-shot cell type prediction and offer insights into its potential benefits in cross-modality prediction and data integration. In summary, we study the application of SSL in SCG, minimizing model bias through simple, fully connected networks, and benchmark SSLs utility across key representation learning scenarios.

bioinformatics↗

bioRxiv2023

Optimal distance metrics for single-cell RNA-seq populations

In single-cell data workflows and modeling, distance metrics are commonly used in loss functions, model evaluation, and subpopulation analysis. However, these metrics behave differently depending on the source of variation, conditions and subpopulations in single-cell expression profiles due to data sparsity and high dimensionality. Thus, the metrics used for downstream tasks in this domain should be carefully selected. We establish a set of benchmarks with three evaluation measures, capturing desirable facets of absolute and relative distance behavior. Based on seven datasets using perturbation as ground truth, we evaluated 16 distance metrics applied to scRNA-seq data and demonstrated their application to three use cases. We find that linear metrics such as mean squared error (MSE) performed best across our three evaluation criteria. Therefore, we recommend the use of MSE for comparing single-cell RNA-seq populations and evaluating gene expression prediction models.

bioinformatics↗

bioRxiv2022

FMSClusterFinder: A new tool for detection and identification of clusters of sequential motifs with varying characteristics inside genomic sequences

This paper describes FMSClusterFinder, a new tool and algorithm for identification and detection of clusters of sequential blocks inside the DNA and RNA subject sequences. Gene expression and genomic groups performance is under the control of functional elements cooperating with each other as clusters. The functional motifs or blocks are often comparably short, degenerate and are located within varying distances from each other. Since functional motifs mostly act in relation to each other as clusters, finding such clusters of blocks is an effective approach to identify functional groups and their function and structure, which represents the need for development of new corresponding tools. The presented web application finds clusters of sequential blocks, with even altering sequences and located in varying distances from each other inside the subject sequences, simultaneously. Additionally, the blocks could be searched with user defined constant or varying characteristics such as: a) different levels of similarity, b) varying minimum number of blocks required to build up the query cluster, c) different types of sequence (degenerate or standard) and d) one or multiple alternative sequences for each block. FMSClusterFinder is available freely at http://fmsclusterfinder.fmsbiog.com.

bioinformatics↗

bioRxiv2021

A Mixed-Modeling Framework for Whole-Brain Dynamic Network Analysis

The emerging area of dynamic brain network analysis has gained considerable attraction in recent years. While current tools have proven useful in providing insight into dynamic patterns of brain networks, development of multivariate statistical frameworks that allow for examining the associations between phenotypic traits and dynamic patterns of system-level properties of the brain, and drawing statistical inference about such associations, has largely lagged behind. To address this need we developed a mixed-modeling framework that allows for assessing the relationship between any desired phenotype and dynamic patterns of whole-brain connectivity and topology. Unlike current tools which largely use data-driven methods, our model-based method enables aligning neuroscientific hypotheses with the analytic approach. We demonstrate the utility of this model in identifying the relationship between fluid intelligence and dynamic brain networks using resting-state fMRI (rfMRI) data from 200 subjects in the Human Connectome Project (HCP) study. To our knowledge, this approach provides the first model-based statistical method for examining dynamic patterns of system-level properties of the brain and their relationships to phenotypic traits.

neuroscience↗

bioRxiv2020

Deep feature extraction of single-cell transcriptomes by generative adversarial network

MotivationSingle-cell RNA-sequencing (scRNA-seq) has opened the opportunities to dissect the heterogeneous cellular composition and interrogate the cell-type-specific gene expression patterns across diverse conditions. However, batch effects such as laboratory conditions and individual-variability hinder their usage in cross-condition design. ResultsWe present single-cell Generative Adversarial Network (scGAN). Our main contribution is to introduce an adversarial network to predict batch effects using the embeddings from the variational autoencoder network, which does not only need to maximize the Negative Binomial data likelihood of the raw scRNA-seq counts but also minimize the correlation between the latent embeddings and the batch effects. We demonstrate scGAN on three public scRNA-seq datasets and show that our method confers superior performance over the state-of-the-art methods in forming clusters of known cell types and identifying known psychiatric genes that are associated with major depressive disorder. AvailabilityThe code is available at https://github.com/li-lab-mcgill/singlecell-deepfeature Contactyueli@cs.mcgill.ca

bioinformatics↗