Introduction to Regulatory Genomics and Epigenomics II: Motif Finding
summary
Covers practical algorithms and data: MEME EM motif discovery uses an EM formulation to learn PWMs from unlabeled l‑mers; a simple k‑mer scoring method averages intensities across ~41,000 36‑bp PBM probes to yield reliable 8‑mer scores; BEAM-ML PBM fits a biophysical energy model and competes with REP in accuracy. Reports k‑mer models boost in vitro AUC (~88→~93) but lose power in vivo, and develops a bidirected de Bruijn construction that provably halves PBM probe count with a polynomial‑time optimal algorithm.
AI-written summary · about
chapters (13)
- 0:00Discovering binding site motifs
- 2:15Binding preferences
- 9:29TFBS models - PWM - Position weight matrix
- 16:27Motif discovery: Goals and challenges
- 20:32EM-reminder: the algorithm Main component
- 31:10The MEME Suite
- 32:22Protein binding microarrays
- 34:56Computational challenge
- 45:08BEEML-PBM model parameters
- 45:49Results
- 48:14The next step: High-throughput SELEX
- 55:14Open challenges
- 56:05Summary
related talks
38mInterpreting Deep Learning Models Of Functional Genomics Data To Decode Regulatory Sequence...
31mWhole Genome Methylation Patterns as a Biomarkers for EHR Imputation
34mModule Discovery in 3-way Data
35mTranscriptional Enhancers - Looking out for the Genes and Each Other
53mApplied Functional Programming in Computational Genomics
31mHacking the Human Genome Using Clojure and Similarity Search
32mMulti-Omic Integration for Understanding Disease
28mNanopore Basecalling for Directed Evolution