M. Gerstung

Teacher-student collaborated multiple instance learning for pan-cancer PDL1 expression prediction from histopathology slides

The paper develops MILTS, a teacher-student framework for predicting PDL1 expression from H&E slides using weakly supervised learning, achieving high AUC scores across multiple tumor types.

Epithelioids: Self-sustaining 3D epithelial cultures to study long-term processes

This paper introduces epithelioids, a 3D culture system for mouse and human epithelia that maintains structure and function over long periods, enabling study of long-term biological processes.

p53 mutation in normal esophagus promotes multiple stages of carcinogenesis but is constrained by clonal competition

This study shows p53 mutations in normal esophagus drive clonal expansion but are constrained by clonal competition, and LOH is critical for malignant transformation.

The longitudinal dynamics and natural history of clonal haematopoiesis

This study tracks clonal haematopoiesis clones over time, finding that mutation type influences growth rates and malignant risk.

Precancer: Mutant clones in normal epithelium outcompete and eliminate esophageal micro-tumors

The paper shows that mutant clones in normal epithelium outcompete and eliminate esophageal micro-tumors through cell competition, preventing tumor formation.

Notch1 mutation drives clonal expansion in normal esophageal epithelium but impairs tumor growth

The paper shows NOTCH1 mutations promote clonal expansion in normal esophageal epithelium but hinder tumor growth, suggesting NOTCH1 inhibition could prevent esophageal squamous cancer.

Spatial genomics maps the structure, character and evolution of cancer clones

The paper develops a spatial genomics method to map cancer clone structure, character, and evolution across breast cancer progression stages.

Characterizing genetic intra-tumor heterogeneity across 2,658 human cancer genomes

This study characterizes intra-tumor heterogeneity across 2,658 human cancer genomes, identifying subclonal mutations and evolutionary patterns across 38 cancer types.

Biallelic mutations in cancer genomes reveal local mutational determinants

The paper identifies biallelic mutations in cancer genomes and highlights their implications for mutation hotspots and variant calling.

Learning mutational signatures and their multidimensional genomic properties with TensorSignatures

This paper introduces TensorSignatures, an algorithm that jointly learns mutational signatures across mutation types and genomic context, revealing new insights into mutational processes in cancer.

The evolutionary history of 2,658 cancers

This study uses whole-genome sequencing of cancer samples to reconstruct the evolutionary history of 39 cancer types, identifying early and late-stage mutational patterns and driver genes.

Genomic landscape and chronological reconstruction of driver events in multiple myeloma

This study reconstructs the genomic evolution of multiple myeloma by analyzing 67 genomes from 30 patients and public datasets, identifying 7 subgroups defined by distinct driver mutations and chronological patterns of genomic events.

Neutral tumor evolution?

The paper explores tumor evolution through subclonal analysis and discusses the role of driver and passenger mutations in cancer progression.