Module Discovery in 3-way Data
summary
Probabilistic model for three-way data (subjects × rows × time) that finds a core gene/parcel set plus subject-specific rows and subject-specific time intervals, implemented as the twigs algorithm: initialize with bi-clustering (biMax / x-marks), then refine by Gibbs sampling and mask discovered modules (x marks + Gibbs + masker). Simulations plant 50 modules (500 rows, 10 subjects) and recovery improves over prior methods; real results include sepsis (14 patients, 5 timepoints) modules enriched for bacterial response and T‑cell activity, and fMRI (20 subjects, 500 parcels, 94 timepoints) modules matching resting-state networks.
AI-written summary · about
chapters (17)
- 0:00Intro
- 0:53A module map
- 4:53The winning alg
- 7:06Outline
- 7:51Three-way data
- 13:19Modules in 3-way data Core module
- 15:02The generative model: scheme
- 16:19The posterior
- 18:11The Algorithm
- 19:31Finding multiple modules
- 20:45Simulation results
- 22:10Gene Expression analysis
- 23:20Core module 1: non-core enrichments
- 24:37Brain fMRI analysis
- 26:52Brain fMRI vs. GE results
- 27:41Comparison to other methods
- 28:05Enrichment tests (GE)
related talks
01h 06mIntroduction to Biological Network Analysis III: Identifying Network Modules
28mSome Current Computational Challenges in Bioinformatics
01h 20mIntroduction to Regulatory Genomics and Epigenomics II: Motif Finding
20mDiagnostic Medical application of ML
36mFormatting Biological Big Data to Enable (Personalized) Systems Pharmacology
30mLeveraging Molecular Data for Drug Discovery
33mUnderstanding Molecular Complexity for Precision Medicine
31mHacking the Human Genome Using Clojure and Similarity Search