project-idea

TLS-Scope: An Open-Source Spatial Transcriptomics Pipeline for Tertiary Lymphoid Structure Classification

A hierarchical graph neural network operating on spatial transcriptomics data can classify tertiary lymphoid structures as immunogenic or tolerogenic — a distinction with direct implications for predicting immunotherapy response that bulk RNA analysis cannot make.

The AI model described in this paper solves a clinically critical problem, but it will only have impact if it's accessible to the research community. TLS-Scope would be an open-source Python pipeline that takes 10x Visium spatial transcriptomics data as input and outputs TLS classification (immunogenic vs. tolerogenic), along with annotated spatial maps showing the location and type of each TLS in the tissue.

The pipeline would include: spot-level cell type deconvolution (to identify immune cell compositions), TLS detection (identifying spatial clusters with TLS architecture), the hierarchical GNN classifier (adapted from the paper's approach), and a visualization module that overlays TLS classification onto the tissue image. All components would be modular so researchers can swap in alternative cell type deconvolution or GNN architectures.

A secondary application: using TLS classification scores alongside existing clinical data to train and validate an immunotherapy response predictor across public cancer datasets. This could be packaged as a companion tool that outputs a probability score alongside the TLS classification — giving oncologists a data-rich immunotherapy response prediction from a single spatial transcriptomics run.

Who Is This For?

Cancer immunologists, bioinformaticians, and computational pathologists working on tumor microenvironment characterization and immunotherapy biomarkers.

Skills & Tools Needed

Python (PyTorch Geometric or DGL for GNN)
Single-cell/spatial transcriptomics analysis (Scanpy, Squidpy)
Deep learning for graph data
Spatial data visualization
Cancer immunology knowledge

Feasibility

medium — Spatial transcriptomics GNN tools exist as frameworks, but implementing a validated TLS-specific classifier requires access to labeled training data and expertise in both spatial biology and deep learning.

Inspired by: A Hierarchical Spatial Graph Neural Network Resolves Immunogenic and Tolerogenic Tertiary Lymphoid Structures in Renal Cell Carcinoma

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.