Identifying Disease Mechanisms Using Gene Networks as Priors

on YouTube ↗39m

" summary

A probabilistic graphical model combines per-variant harmfulness priors, aggregated coding scores and gene-level expression aberration scores, then ties genes together with a degree-corrected network prior implemented via loopy belief propagation. Simulations (mechanisms of 10–100 genes, varying sample size and noisy edges) show higher precision and robustness than greedy network methods that use precomputed p-values (dmGWAS/SKAT comparisons). Applied to TCGA breast cancer the pipeline estimates coding variants explain ~55% of cases, coding+regulatory joint hits ~15% and regulatory-only ~20%, and it highlights patient-level posterior gene probabilities in medulloblastoma cohorts.

AI-written summary · about

" chapters (22)

1. 0:00Intro
2. 0:56Identifying genetic disease mechanisms
3. 2:11Data that can be used to identify the mechanism
4. 3:02Aggregation based methods
5. 3:43Network based methods
6. 4:12List methods
7. 7:02Model of a genetic disease
8. 8:53Method Overview
9. 11:33Graphical model
10. 12:23Network implementation
11. 14:05Overcoming network complexities
12. 16:51Key differences
13. 18:21Simulation data
14. 19:28Results: Simulated mechanism of 10 genes
15. 23:18Results: sensitivity to the noise
16. 24:15Robustness of the method
17. 25:07Advantages
18. 25:47Real data: TCGA breast cancer
19. 26:27Comparing to other methods
20. 28:27Luminal breast cancer findings
21. 29:32Medulloblastoma
22. 30:16SHH Predictions based on model

" related talks

• 56mIntroduction to Cancer Bioinformatics III: Cancer Progression NetworksNiko Beerenwinkel · Algorithmic Challenges in Genomics Boot Camp
• 01h 19mAnalyzing Genetic Networks Using Neural NetworksManu Flores · PyData Los Angeles 2019
• 33mUnderstanding Molecular Complexity for Precision MedicineNatasa Przulj · From Algorithms to Discovery in Genome-Scale Biology and Medicine 2022
• 01h 06mIntroduction to Biological Network Analysis III: Identifying Network ModulesRoded Sharan · Algorithmic Challenges in Genomics Boot Camp
• 17mLeveraging Knowledge Graph and Imaging to Predict Gene–Disease Links in Cardiovascular DiseasesKhaled Rjoob · NODES 2025
• 01h 31mIntroduction to Cancer Bioinformatics I: Inferring Genomic Variation from Tumor Sequencing DataBen Raphael, Niko Beerenwinkel · Algorithmic Challenges in Genomics Boot Camp
• 38mMixed and Covariate-Dependent Graphical ModelsLiza Levina · Workshop on Succinct Data Representations and Applications (Simons Institute Big Data 2013)
• 30mLeveraging Molecular Data for Drug DiscoveryHaiyan Huang · Computational Challenges in Very Large-Scale 'Omics'