Author: Gireesh Bogu
Date: Dec 1st 2013
Email: gireesh.bogu@crg.eu
http://mcb.asm.org/content/36/5/809.abstract
First, I collected mapped ChIP-Seq reads of H3K4me1, H3K4me3, H3K36me3, H3K27me3, H3K27ac, CTCF and RNA polymerase II from ENCODE http://hgdownload.cse.ucsc.edu/goldenPath/mm9/encodeDCC/wgEncodeLicrHistone/. This data was originally produced from mouse (C57BL/6-strain, E14 or 8 week-old) brain, heart, kidney, liver, small intestine, spleen, testes, thymus and embryonic stem (ES) cell line as a part of mouse ENCODE project.
Second, I used a Poisson-based multivariate hidden Markov model (ChromHMM) to identify regions or states enriched in specific combinations of histone modifications as described previously but without extension of the reads.
Third, after testing models from a 2-state model to a 50-state model, we decided to use the 15-state model, which allowed me to interpret the chromatin frequency observed across various tissues and cell lines.
This resulted in six major chromatin states of active promoter, poised promoter, strong enhancer, poised or weak enhancer, insulator, repressed, transcribed and heterochromatin states. In total, 3,612,616 regions in the mouse genome were enriched with at least one of the six major chromatin states.
These were the following commands that were used to generate maps of chromatin states in mouse
-
java -jar ChromHMM.jar BinarizeBed CHROMSIZES/mm9.txt mm9EncodeHMM/ mm9EncodeHMM/cellmarkfile mm9EncodeHMM_BinaryOutput/
-
java -jar ChromHMM.jar LearnModel mm9EncodeHMM_BinaryOutput/ mm9EncodeHMM_Output15/ 15 mm9
To interpret states, you can use this figure where I named each state with a specific color and name (for a detailed description please look at my work):
Note: If you are using the chromatin state maps from here, try to use the chromatin state names as reference but not the numbers.
-
A map of the cis-regulatory sequences in the mouse genome, Nature, 2012, Yin Shen, Feng Yue, David F. McCleary, Zhen Ye, Lee Edsall, Samantha Kuan, Ulrich Wagner, Jesse Dixon, Leonard Lee, Victor V. Lobanenkov & Bing Ren
Data source: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE29184
-
ChromHMM: automating chromatin-state discovery and characterization, Nature methods, 2012, Jason Ernst & Manolis Kellis
Software link: http://compbio.mit.edu/ChromHMM/
-
Making sense of chromatin states, Nature, Monya Baker