ChromHMM annotations predicted from GECSI from Jason Ernst Lab (jason.ernst@ucla.edu) (Jingyuan Fu and Jason Ernst)
=======================================================

GECSI-predicted per-epigenome chromatin state annotations on 414 epigenomes from IHEC EpiAtlas that have fully observed experimental data of H3K4me1, H3K4me3, H3K9ac, H3K27ac, H3K27me3, H3K36me3 and RNA-seq. GECSI's predictions are generated using a 5-fold cross validation.

File ./bed/*.bed.gz - per-epigenome chromatin state annotations in a 9 column bed format (chrom, chromStart, chromEnd, name, score, strand, thickStart, thickEnd, itemRgb). A checksum file is included in this directory. Tracks are named in the following format: GECSI_[IHEC_ID]_[harmonized_sample_label].bed.gz, where “IHEC_ID” refers to the unique EpiRR ID assigned to each IHEC sample, and “harmonized sample label” refers to the corresponding column as represented in the IHEC epigenome v1.3 metadata. 
File ./bigbed/*.bb - per-epigenome chromatin state annotations in a bigbed format. A checksum file is included in this directory. Tracks are named in the following format: GECSI_[IHEC_ID]_[harmonized_sample_label].bed.gz, where “IHEC_ID” refers to the unique EpiRR ID assigned to each IHEC sample, and “harmonized sample label” refers to the corresponding column as represented in the IHEC epigenome v1.3 metadata. 
File ./posterior/*.txt.gz - per-epigenome chromatin state posterior probabilities in txt format. A checksum file is included in this directory.