Elham Azizi

The paper uses in vivo single-cell CRISPR screens to study how tumor suppressor loss and genetic perturbations affect glioma states and immune recruitment in mice.

SHERLOCK is a deep generative framework for single-cell perturbation analysis that learns interpretable perturbation representations and enables causal analysis of downstream effects.

The paper identifies a T-cell receptor structural signature that predicts acute GVHD risk by analyzing alloreactive clonotypes and introduces a metric to quantify clonal dominance.

LYNX is a deep generative model that integrates multimodal spatial data to model spatial dynamics and cell-cell interactions across tissues.

ARCADIA integrates scRNA-seq and spatial proteomics to reveal spatially dependent transcriptional programs without cell barcode pairing or feature-level correspondence.

The paper uses multiplex imaging to track the development of T cell-APC triad neighborhoods during immune checkpoint therapy, showing their role in spatially organizing tumor immunity.

This paper develops a continual learning framework to map colorectal cancer cell states to causal mechanisms through integration of single-cell data and perturbation experiments.

The paper develops a CRISPR screening framework to map kinase-dependent immune evasion mechanisms in glioblastoma by integrating single-cell transcriptomics and deep generative models.

The paper identifies co-expression patterns of chemokine receptors on CD8+ T cells that distinguish anti-tumor efficacy from off-target toxicity in melanoma.

Squidiff predicts transcriptomic changes across cell types in response to environmental stimuli using a diffusion model, enabling in silico screening for drug discovery and precision medicine.

MrVI is a deep generative model that identifies sample-level stratifications and cellular differences without prior cell-type grouping, applied to COVID-19 and IBD data.

The paper improves patient-level phenotype prediction from scRNA-seq data by learning domain-invariant features, showing better out-of-domain accuracy and biological fidelity.

AMICI estimates cell-cell interaction length scales and links communication to gene programs in spatial transcriptomic data.

Decipher is a deep generative model that integrates single-cell RNA-seq data to visualize and characterize transitions between normal and perturbed cell states across different diseases.

This study identifies ASXL1 truncating mutations as a driver of leukemic resistance to T cell attack and proposes EZH2 inhibition as a therapeutic strategy to overcome this resistance.

This study uses spatiotemporal single-cell analysis to track T cell clonal dynamics and phenotypic plasticity in GVHD, revealing links between TCR features, clonal expansion, and disease severity.

The paper analyzes single-cell data from bone marrow samples to identify immune cell interactions that predict response to adoptive cellular therapy in myeloid leukemia.

Echidna quantifies gene dosage effects on phenotypic plasticity by integrating scRNA-seq and WGS data to separate genetic from cell non-autonomous mechanisms.

This paper introduces RefCM, a method that improves cell type annotation in scRNA-seq data using optimal transport and integer programming, achieving accurate cross-technology, cross-tissue, and cross-species mappings.

DIISCO infers dynamic intercellular interactions from time-series single-cell data using a Bayesian framework, showing its ability to capture time-dependent communication patterns in simulated and CAR-T cell data.

This study uses SCHEMATIC to analyze the molecular response of glioblastoma to EGFR inhibitors, identifying chemotype-specific programs and showing that combined EGFR/PI3K inhibition reduces adaptive transcription.

CRISPRmap enables multi-omic optical pooled screens to map gene function in cells and tissue without sequencing, identifying pathogenic mutations and analyzing spatial phenotypes.

CellStitch improves 3D cell segmentation in anisotropic images using optimal transport, outperforming existing methods on anisotropic datasets.

Starfysh integrates spatial transcriptomics and histology to analyze tumor microenvironments, identifying disease-specific spatial hubs and immune interactions in breast cancers.

KINOMO improves the analysis of single-cell RNA-seq data by recovering intra- and inter-tumoral heterogeneity through non-negative matrix factorization, outperforming existing methods in benchmarking.

This study develops a method for single-cell and whole-genome sequencing of frozen clinical specimens, demonstrating its utility in analyzing tumor evolution and rare disease biospecimens.

The paper maps T cell state evolution during response and resistance to adoptive cellular therapy using single-cell transcriptomics and machine learning, identifying key exhausted subsets and their origins.

The paper introduces a transfer learning framework for cell segmentation that incorporates geometric features to improve performance on in situ tissue images using in vitro data.