verion 3 update

Author: havu73

README.md

@@ -34,7 +34,7 @@ chrom, start, end, full_stack state
 - An analogous annotation in mouse can be found on <a href="https://github.com/ernstlab/mouse_fullStack_annotations">this page</a>.
 # Notes about the versions of full-stack annotation:
 Since the publication of our data, the annotation of full-stack annotation in **hg19** have **NOT** been changed (__except for our state names being changed from 0-based indexing system 0_GapArtf1-99_TSS2 to 1_GapArtf1-100_TSS2__). The **hg38** annotation have been changed in 3 version**: 
-- Version 1 (from publication of this repository to August 11 2023): In this version, we tried to liftOver the annotation from hg19 to hg38 such that no bp in hg38 were mapped from multiple bps in hg19. However, it was brought to our attention in February 2024 that there are still some regions in hg38 that were mapped from multiple regions in hg19 in this version of the annotation. This is caused by an implicit assumption in our code that each 200-bp segment from hg19 will be mapped to another segment of length 200-bp in hg38 by the liftOver tool provided by UCSC genome browser. This is not the case, so in version 3 (May 6 2024 onwards), we fixed this issue with updated code in the </a href="https://github.com/ernstlab/full_stack_ChromHMM_annotations/tree/main/chromHMM_utilities/liftOver">liftOver pipeline</a>. In this first version, the states were numbered by the 0-based system (0_GapArtf1-99_TSS2, but note that the names GapArtf1, TSS2, etc. are exactly how they appeared in the paper). In the current version, we fixed this issue and eliminated 2,979,514 bp in verion1’s annotation that were wrongly mapped from different regions in hg19 (0.09% of the hg38 assembly genome). 
+- Version 1 (from publication of this repository to August 11 2023): In this version, we tried to liftOver the annotation from hg19 to hg38 such that no bp in hg38 were mapped from multiple bps in hg19. However, it was brought to our attention in February 2024 that there are still some regions in hg38 that were mapped from multiple regions in hg19 in this version of the annotation. This is caused by an implicit assumption in our code that each 200-bp segment from hg19 will be mapped to another segment of length 200-bp in hg38 by the liftOver tool provided by UCSC genome browser. This is not the case, so in version 3 (May 6 2024 onwards), we fixed this issue with updated code in the </a href="https://github.com/ernstlab/full_stack_ChromHMM_annotations/tree/main/chromHMM_utilities/liftOver"> liftOver pipeline</a>. In this first version, the states were numbered by the 0-based system (0_GapArtf1-99_TSS2, but note that the names GapArtf1, TSS2, etc. are exactly how they appeared in the paper). In the current version, we fixed this issue and eliminated 2,979,514 bp in verion1’s annotation that were wrongly mapped from different regions in hg19 (0.09% of the hg38 assembly genome). 
 - Version 3 (May 6 2024 onwards): In this version, the state annotation for hg38 is corrected such that there are absolutely no bps in hg38 that were mapped from multiple bps in hg19. The states are numbered by the 1-based system (1_GapArtf1-100_TSS2, but note that the names GapArtf1, TSS2, etc. are exactly how they appeared in the paper). The change from 0-based to 1-based state numbering was made for consistency with other universal chromatin state model annotations. 
 - Version 2: (From August 11 2023___ to February 26 2024___ ) : We mistakenly shared the links for the data in this version such that the annotations hg38 are raw output of the UCSC Genome browser's liftOver software (missing the step of filtering out bps in hg38 that were mapped from multiple bps in hg19). 
 As of May 6 2024, we have archived the data of version3 onto https://zenodo.org/records/5759119