Modeling Mutational Patterns in Tumor Genomic Data

on YouTube ↗28m

" summary

Proposes a muX generative model that jointly models coverage, impurity and false positives/negatives and a waiting-time model that explains mutual exclusivity by early fixation of one pathway mutation; significance uses likelihood-ratio tests. Builds an SL scoring pipeline that ranks pairs by mutual exclusivity plus survival difference (Kaplan–Meier AUCs) and tests with Cox models; benchmarks against ~7,000 experimentally assayed pairs (few true SLs, low AUC), and extends predictions using PPI motifs.

AI-written summary · about

" chapters (9)

1. 0:00Intro
2. 0:16Tumors of different patients are heterogeneous
3. 1:10How to make sense from the jungle of mutations?
4. 2:47What are mutually exclusive patterns of cancer mutations?
5. 3:53Why cancer pathways → mutual exclusivity?
6. 6:51Motivation for muex
7. 12:51Harnessing synthetic lethality (SL) in cancer treatment
8. 16:40Detecting SL gene pairs from genomic and clinical data
9. 21:32How well can we do in SL prediction?