Inferring Evolution by Copy Number Variations in Tumor Cell Populations

on YouTube ↗35m

" summary

Uses FISH with 8–15 probes on hundreds of single cells per tumor to record copy number gains and losses, models evolution as probe-wise gains/losses mapped to a rectilinear minimum Steiner tree and prunes search space with median-joining/neighbour-joining heuristics, then extends to multiscale events (whole-genome duplication) and patient-specific rate learning via EM. Finds tumor phylogeny topology and learned parameters outperform raw data for classification—roughly ~80% classification accuracy for primary versus progressing tumors—and emphasizes pervasive random diversification from a chromosome instability mutator phenotype.

AI-written summary · about

" chapters (15)

1. 0:00Intro
2. 0:21Cancer Progression is an Evolutionary Phenomenon (Nowell. Science, 194:23-28, 1976)
3. 2:49Tracking Tumor Evolution with Fluorescence in Situ Hybridization (FISH) Copy Number Variation (CNV) Data
4. 4:46Phylogenetics of Tumor Cell Populations by Fluorescence in Situ Hybridization (FISH)
5. 5:58Most Tumors Show Extensive Cellular Heterogeneity Heterogmely from paired ductal carcinoma in situ (DOS) and invasive ductal carcinoma (IDC)
6. 8:42Phylogenetics of Copy Number Variations
7. 12:46Gene Gain/Loss Statistics from Tumor Phylogenies across Progression Stages
8. 13:29Topological Variation in Primary vs. Metastatic Cervical Cancer Phylogenies
9. 15:08Survival Prediction for 65 Head-and-Neck Cancers Disease-free survival
10. 16:05Cervical Tumor Classification from Tree Statistics
11. 17:48Improving the Model: Multiscale CNV Evolution
12. 20:04Tumor Evolution Involves Complex Diversification and Selection
13. 21:08Inferred Evolutionary Rate Parameters Breast Cancer
14. 21:43Weighted Models Lead to Substantial Changes in Tree Topology
15. 23:23Hyperbolic, Overstated Conclusions