Dive into the world of epigenomics and Hidden Markov Models (HMMs) in this comprehensive lecture. Explore the three types of epigenomic modifications, including histone modifications, DNA methylation, and chromatin accessibility. Learn about diverse histone modifications, methylation profiling techniques, and data generation methods like ChIP-Seq. Discover read mapping algorithms, quality control processes, and peak calling techniques. Delve into the foundations of HMMs, including their parameters, parsing, and decoding methods. Understand the Viterbi algorithm for finding the best parse and posterior decoding for determining the most likely state. Gain insights into non-standard modifications, epigenetic inheritance, and developmental memory establishment through engaging Q&A segments.
Epigenomics and Hidden Markov Models in Computational Biology - Lecture 5
Overview
Syllabus
Logistics
Lecture Overview
Introduction to Epigenomics
Three types of Epigenomic Modifications
Q1: Non-standard modifications
Q2: Epigenetic inheritance
Q3: Developmental memory establishment
Diversity of Histone modifications
Methylation Bisulfite and DNase Profiling
Antibodies, ChIP-Seq, data generation projects, raw data
Read mapping: Hashing, Suffix Trees, Burrows-Wheeler Transform
Quality Control, Cross-correlation, Peak calling, IDR similar to FDR
Discovery and characterization of chromatin states
HMM Foundations, Generating, Parsing, Decoding, Learning
Two Sets of HMM parameters: Emissions, Transitions
Example 2-state HMM, observations, scoring, inference
Viterbi algorithm: Find best parse π*= argmaxπPx,π
Posterior Decoding: Most likely state πiover all paths
Summary
Taught by
Manolis Kellis
