Jeffrey H. Chuang

The paper analyzes the spatial molecular architecture of melanoma tumors under different immunotherapy regimens and identifies biomarkers predicting response to treatment.

DIANNE enables rapid training and inference of spatial differential attributes in histology images without manual annotations, using mixup augmentation and foundation models.

The paper identifies two distinct senescent β-cell subpopulations with different functional roles in islet aging and dysfunction using single-cell proteomics and transcriptomics.

The paper introduces SparTile, a computational method for analyzing multiplex proteomics images to identify clinically relevant spatial microenvironments in TNBC, finding specific markers and distances associated with poor survival.

This study investigates cohesin's role in transcriptional loops and DNA replication using human cells, finding that cohesin loss affects long-range chromatin interactions and up-regulates early initiation zone genes.

SpliceDecoder prioritizes biologically meaningful splicing events by predicting their impact on transcript productivity, protein sequence, and functional domains.

The paper identifies senescent β-cell subpopulations in humans and mice that induce pathologic dysfunction through paracrine signaling and shows JAK inhibitors can reverse this effect.

The paper develops a Nextflow pipeline for processing Visium spatial transcriptomics and H&E images from PDX samples, enabling multi-modal analysis and integration of spatial gene expression and morphometric features.

This study uses spatial transcriptomics and deep learning to analyze melanoma treatment resistance, identifying persister states and resistance mechanisms with spatiotemporal resolution.

Cellos is a high-throughput image analysis pipeline for 3D organoid segmentation and analysis, enabling accurate quantification of drug responses and cell interactions in cancer organoids.

This study creates a single-cell aging atlas of mouse mammary tissues, identifying shared epigenomic and transcriptomic changes between aging and cancer.

SAMPLER provides a statistical method to generate efficient whole slide image representations, achieving comparable or better accuracy than attention models while being significantly faster.

The paper presents a pan-cancer PDX histology image repository with genomic and pathological annotations for deep learning analysis of tumor regions and xenograft-transplant lymphoproliferative disorder prediction.

This paper presents a CNN that predicts HER2 status and trastuzumab response from H&E images with higher accuracy than IHC.

This paper introduces 'mones' as interpretable CNN features from H&E images that correlate with biological phenotypes and can accurately classify tumors and normals.

This paper uses CNNs to classify and compare histopathological images across cancer types, revealing conserved spatial patterns and mutation-related features.

This study shows that GTPBP1 resolves ribosome pausing caused by tRNA deficiency in the mouse brain, preventing neurodegeneration and highlighting the role of translational GTPases in neuronal homeostasis.

pyBedGraph is a Python package that rapidly computes summary statistics for genomic signal tracks, outperforming existing tools by up to 245 times.

The paper introduces MIA-Sig, an algorithm to de-noise chromatin interaction data and identify topological domains and inter-domain contacts.

The paper develops bioinformatics workflows to analyze genomic data from patient-derived xenografts, addressing challenges like separating mouse and human reads and identifying somatic mutations without paired normal DNA.

The paper evaluates the robustness of mutation calling algorithms on tumor heterogeneity measures and finds limited clinical predictive power for MATH score.

SARNAclust identifies RNA motifs from immunoprecipitation data by combining structure/sequence analysis, revealing known and novel motifs including a new ILF3 motif similar to hnRNPC.

This study quantifies intratumoral subclone dynamics in breast cancer xenografts, revealing a treatable ecology where subclone ratios predict treatment response.