Sohrab P. Shah

The paper shows that error-correction objectives for perfect phylogenies are statistically inconsistent, often favoring incorrect trees over the true one in simulated data.

scPlOver infers DNA content from single-cell WGS data using fragment overlaps, improving accuracy in identifying high DNA content cells in ovarian cancer.

The paper analyzes the evolutionary trajectories of high-level copy-number amplifications in cancer cell genomes to distinguish between extrachromosomal and intrachromosomal amplification mechanisms.

This study introduces iMSEA, a computational framework that identifies metabolic state shifts in BRCA1-deficient cancers using omics data and validates findings with experimental assays.

This study tracks clonal evolution of drug resistance in ovarian cancer using structural variants in cfDNA, finding polyclonal resistance with dominant high-fitness clones and genomic features linked to drug resistance.

The paper introduces ProteomeGenerator3, an end-to-end proteogenomics method that detects non-canonical proteoforms in cancer using long-read transcriptomics and multi-dimensional proteomics.

This study shows that ongoing whole-genome doubling in ovarian cancer promotes cell diversity and immune dysregulation through chromosomal instability and quiescent states.

This study identifies MAPK activation and cellular plasticity as complementary resistance mechanisms to EGFR TKI in lung adenocarcinoma using single-nuclei RNA sequencing and clinical data.

The paper identifies impaired T cell priming and KLF2-dependent quiescence in gastric cancer draining lymph nodes as mechanisms limiting immune responses.

This paper develops a multimodal transformer model using H&E images and text reports to predict recurrence scores in hormone receptor-positive breast cancer, achieving high accuracy and generalizability.

This paper introduces ArtiCull, a method to analyze single-cell whole-genome sequencing data to distinguish active from historical mutational processes in cancer, revealing temporal and spatial patterns in mutational signatures.

The paper analyzes copy number alterations in breast epithelial cells from BRCA mutation carriers and wildtype individuals, finding lineage-specific CNAs that mirror cancer genomes.

The paper identifies genomic and non-genomic contributions to drug resistance in triple negative breast cancer using single cell transcriptomics and copy number data.

The paper identifies how germline and somatic mutations interact to cause resistance to CDK4/6 inhibitors in breast cancer and proposes a sequential treatment strategy using PARP inhibitors.

The paper introduces a scalable Bayesian method for phylogenetic tree inference from single-cell whole genome sequencing data, demonstrating its application on three datasets.

GeneVector uses mutual information-based dimensionality reduction to identify transcriptional programs and classify cell types across single-cell RNA-seq datasets.

This paper introduces TCRi, an information theoretic framework to analyze TCR and gene expression data from cancer patients, showing improved characterization of clonotypes and their phenotypic changes post-therapy.

Epiclomal clusters sparse single-cell DNA methylation data and imputes missing values, outperforming non-probabilistic methods and revealing sub-clonal methylation patterns in tumour genomes.

This study identifies transcriptional stress responses in single-cell RNA-seq data from dissociated tumour tissues, showing that collagenase digestion induces a conserved stress response minimized by cold active protease.

clonealign integrates single-cell RNA and DNA sequencing data from cancer samples to identify clone-specific gene expression patterns and dysregulated pathways.

Project Dhaka uses a variational autoencoder to analyze single-cell genomic data, revealing tumor subpopulations and their evolutionary relationships that other methods miss.

CellAssign probabilistically assigns cell types to scRNA-seq data, revealing spatiotemporal dynamics in human cancers like ovarian and follicular lymphoma.

This study identifies somatic coding and regulatory variants in DLBCL, revealing novel mechanisms of oncogene deregulation and potential therapeutic targets.

The paper introduces scvis, a deep generative model for dimensionality reduction of single-cell RNA-seq data, preserving both local and global structures in low-dimensional representations.

This study introduces a machine learning method to infer mutation signatures from both single nucleotide and structural variations in DNA repair deficient cancers, identifying new age-related and HRD-associated signatures in breast and ovarian cancers.

The paper analyzes the interaction between tumor and immune clones in ovarian cancer, finding that immune contexture patterns mirror malignant spread and that foldback inversions correlate with poor outcomes.

MuClone improves somatic mutation detection and classification by integrating clonal structure information across multiple tumor samples.

The paper defines eIF4A3's role in splicing and NMD through a systems analysis using novel small molecule inhibitors and shows its link to cell cycle regulation.

deStruct improves breakpoint prediction by using specific realignment and clustering, achieving higher sensitivity and specificity than existing tools on simulated data.