project-idea

StromalAtlas PDAC: An Interactive Single-Cell Fibroblast Subtype Browser for Pancreatic Cancer

A single-cell RNA-seq study of 42 PDAC tumors identified an NFATC2-expressing CAF subpopulation associated with better survival — demonstrating that CAF heterogeneity in pancreatic cancer carries prognostic information that bulk analysis misses.

The cancer-associated fibroblast field is exploding with single-cell sequencing data, but it's scattered across dozens of papers with inconsistent naming conventions and no unified way to browse CAF subtypes across studies. StromalAtlas PDAC would be an interactive web browser aggregating published PDAC single-cell datasets, with a focus on fibroblast heterogeneity and its clinical correlates.

The tool would allow researchers to query: 'Show me all fibroblast clusters from PDAC single-cell studies' and then cross-reference these clusters with survival data, treatment response data, and known marker genes. Users could look up NFATC2 (or any gene of interest) and see how its expression in fibroblast clusters correlates with outcomes across multiple cohorts — not just the single dataset from any one paper.

This kind of cross-study integration would rapidly validate (or refute) newly reported CAF subtypes and help the field converge on a unified taxonomy of pancreatic cancer fibroblasts. It would also be a valuable resource for researchers designing anti-stromal therapeutic strategies.

Who Is This For?

Cancer biologists studying the tumor microenvironment, bioinformaticians, and translational pancreatic cancer researchers.

Skills & Tools Needed

Single-cell RNA-seq analysis (Seurat/Scanpy)
Database design and curation
Interactive data visualization (D3.js, Plotly, or similar)
Public data integration (GEO, CELLxGENE)
Web development

Feasibility

medium — Public single-cell PDAC datasets are available, but harmonizing cell type annotations across studies requires significant manual curation and computational expertise.

Inspired by: NFATC2 in pancreatic cancer-associated fibroblasts predicts treatment response and facilitates ERBB-targeted therapies

Landmark DCIS trial, CRC microbiome & AML combos

Two themes cut across this week’s 236 posts. The first is de-escalation: researchers are building rigorous evidence that patients can safely do less — skip lymph node surgery, forgo radiotherapy, watch pre-invasive lesions instead of cutting them out. The landmark result here is a randomized trial confirming that active surveillance

Colorectal Cancer Microbiome Discovery Platform

Analysis of 9,000+ patients revealed that only colorectal cancer — not other cancer types — consistently harbors distinct microbial communities, suggesting the gut microbiome has a unique relationship with colorectal carcinogenesis that could be exploited for diagnostics and treatment.

GBM Tumor Microenvironment Drug Target Atlas

Non-cancerous brain support cells were found to promote glioblastoma growth through paracrine signaling, and blocking this communication dramatically slowed tumor growth — revealing that GBM's microenvironment contains targetable vulnerabilities that are being systematically overlooked by drug…

BET Inhibitor Target Selectivity Database

BRD2 and BRD4 have fundamentally distinct functions — BRD2 prepares genes for activation while BRD4 triggers transcription — meaning that pan-BET inhibitors have been disrupting gene regulation in counterproductive ways, and the field now needs selective BRD4 inhibitors.