Mehrtash Babadi

The paper introduces SOCAM, a supervised cell annotation method that enforces ontological consistency and improves performance using a probability propagation strategy and logistic regression model.

multiVIB integrates single-cell multi-omics data across modalities and species with a probabilistic contrastive learning framework, achieving state-of-the-art performance while preserving biological variation.

A masked-expression Transformer trained on 983 paired H&E and spatial transcriptomics slides predicts 38,984 genes from histology without fine-tuning, leading HEST-Bench and improving survival and mutation prediction.

This paper introduces a service for automated cell type annotation in scRNA-seq data using reverse search against a curated database, providing both fine-grained and coarse annotations without relying on predefined markers.

This study creates a single-nucleus and spatial transcriptomic atlas of the COVID-19 liver, revealing cellular and functional disruptions associated with organ damage.

TissueMosaic performs cross-sample differential analysis of spatial transcriptomics data using self-supervised representation learning to improve detection of genes associated with tissue structure changes.

This study maps the transcriptional landscape of the rat cardiovascular system at single-cell resolution, identifying 26 cell types and revealing tissue-specific molecular diversity and communication networks.

CellMincer is a self-supervised deep learning method for denoising voltage imaging data, achieving significant noise reduction and improved functional phenotype separation.

The paper introduces CytoSummaryNet, a contrastive learning method for aggregating single-cell features in image-based profiling, improving mechanism of action prediction by 30-68%.

CellCap models the relationship between cell state and perturbation response in single-cell experiments, revealing novel insights into cellular behavior.

The paper develops a weakly supervised learning method to learn treatment effect representations from imaging data, improving downstream analysis performance.

CellBender removes systematic background noise from droplet-based single-cell data using a deep generative model, improving alignment with established gene expression patterns.

This paper introduces MAS-ISO-seq, a method to boost throughput of long-read RNA sequencing for isoform discovery, showing improved detection of differentially spliced genes and cell clustering in T cells.

GATK-gCNV detects rare copy number variants from exome sequencing data, achieving high recall in benchmarking and generating a reference catalog of rare coding CNVs in the UK Biobank.

This paper introduces SynapseCLR, a contrastive learning method for 3D EM data to extract synapse features from mouse visual cortex, achieving high accuracy in synapse classification and annotation.

This study creates a single-cell and spatial atlas of autopsy tissues from COVID-19 patients, revealing altered cellular responses and potential therapeutic targets.

The paper uses single-nucleus RNA sequencing to define the transcriptional and cellular diversity of the non-failing human heart, identifying 9 major cell types and 20 subclusters.