Mitchell Paul Levesque

CATFISHR integrates transcriptomic and copy-number data to improve malignant cell identification in scRNA-seq, achieving high accuracy across multiple datasets and identifying previously missed malignant populations.

This study identifies the TERRA-NONO axis as a driver of fibroblast reprogramming into cancer-associated fibroblasts, showing its therapeutic potential in multiple cancers.

The paper demonstrates a method to simultaneously measure genome and proteome from the same cells in archival FFPE tissue, applied to melanoma.

The paper describes a protocol for preparing tissue samples for proteome and phosphoproteome analysis using mass spectrometry, including steps for tissue disruption, lipid depletion, and phosphopeptide enrichment.

This study activates the NLRP1 inflammasome in skin equivalents to reveal a novel signaling axis involving IL-1 and IL-36γ in psoriasis.

The paper identifies a pre-invasive malignant keratinocyte population in actinic keratosis, characterized by p63 and PITX1 regulation, and distinguishes early oncogenic programs from invasion effectors in squamous cell carcinoma.

The paper uses in vivo single-cell ribosome profiling to study translational changes in mouse epidermal stem cells during aging, finding cell-type-specific translational reprogramming in aged stem cells.

This study shows that non-muscle myosin II overactivation occurs in dedifferentiating melanomas adapting to BRAF and MEK inhibitors, and targeting NMII with BMi reduces drug-tolerant persister cell survival.

This study identifies B cell-derived extracellular vesicles and miR-99a-5p as key regulators of melanoma response to immunotherapy.

This study identifies immunological differences in ulcerated acral melanoma, showing increased humoral immunity and inflammation markers, and proposes potential therapeutic targets.

This study identifies CNDP1 as a metabolic vulnerability in brain metastasis, showing its inhibition suppresses metastasis and its expression confers metastatic potential.

This study uses deep learning to classify drug-naive and resistant cancer cells based on morphology, achieving high accuracy across multiple cancer types and drug classes.

This paper develops an AI-based classifier to classify melanoma cell phenotypes using morphology, validated by scRNA-Seq, showing morphology correlates with phenotypic states.

This paper introduces a method to predict single-cell perturbation responses using neural optimal transport, showing improved performance over existing methods on drug-treated cells.

This paper introduces a multiplex-DIA workflow that enhances single-cell proteome depth using dimethyl labeling and a reference channel, achieving reliable quantification of nearly 4,000 proteins per cell.

The paper identifies wound fibroblast phases and matrix patterns that define cancer-associated fibroblast subtypes with prognostic relevance in skin cancer.

scROSHI identifies cell types without training data by using pre-existing gene sets and hierarchical relationships, outperforming methods in low-data and high-diversity scenarios.

gExcite is a start-to-end framework for analyzing single-cell gene expression and CITE-seq data, including hashing deconvolution, and is scalable for multiple samples.

SCIM matches single-cell measurements across technologies by learning a technology-invariant latent space and applying bipartite matching, achieving high accuracy in real-world datasets.

The paper analyzes whole-genome sequencing data from melanoma metastases to show that branched evolution is common and that subclonal heterogeneity is prevalent in metastatic melanoma.