Expression Analysis of Tumors Based on Patterns in Alternative Splicing

on YouTube ↗35m

" summary

Presents an empirical Bayes workflow that models exon inclusion as an overdispersed binomial with empirical Bayes by using Efron’s dispersion trick, showing the MLE of dispersion is approximately gamma so conjugate priors give analytic shrinkage. On simulations the estimator controls FDR versus other methods that report ~10% but actually hit ~40% false discoveries, and on TCGA AML/UVM data proportion‑based clustering finds isoform‑switching signals and plate batch effects and improves sensitivity to shared splicing patterns.

AI-written summary · about

" chapters (14)

1. 0:00Start
2. 4:23Standard Gene Analysis
3. 5:58Exon Usage
4. 7:14Empirical Bayes Strategy
5. 9:26Empirical Bayes For Expression Settings
6. 10:07General Empirical Bayes Method
7. 11:14Performance on Simulated data
8. 13:20Real Data (TCGA)
9. 18:08Clustering on Isoforms vs Genes
10. 23:14Recapitulate Brooks et al
11. 23:47Effectively finds Batch effect in LAML
12. 25:04Differences in clusters
13. 25:23Clustering finding differences between large class of isoforms
14. 27:38ACKNOWLEDGMENTS