This repository contains chromosome/contig name mappings between UCSC <-> Ensembl <-> Gencode for a variety of genomes.
The files are named
AAA is a genome and version (e.g., GRCh37) and BBB and CCC are sources (namely, ensembl, UCSC, or gencode). Each file contains two columns. The first is the chromosome name in
BBB and the second that in
CCC. For example, let's suppose we're interested in converting gencode to ensembl chromosome names for GRCh37. We would then look in the
GRCh37_gencode2ensembl.txt file and would see lines such as:
chrX X chrY Y chrM MT GL877870.2 HG1001_PATCH GL877872.1 HG1032_PATCH GL383535.1 HG104_HG975_PATCH JH159133.1 HG1063_PATCH
In this case,
chrX is the gencode name and
X is the equivalent Ensembl name.
It's not always the case that a given chromosome/contig exists in all sources. An example of that is
GRCh38_gencode2ucsc.txt. There, a number of entries exist in gencode that are absent in UCSC. In cases such as this, the second column in a txt file will simply be empty:
KI270937.1 chr3_KI270937v1_alt KI270938.1 chr19_KI270938v1_alt KN196472.1 KN196473.1
There is always a second tab-separated field above, but
KN196473.1 simply don't exist in UCSC. So a script using these files can simply look for columns with values "" to indicate "missing".
Occasionally, e.g., with mm9, UCSC will merge contigs together into an ordered
*_random sequence. This means that an individual entry in UCSC can map to multiple entries in Ensembl and Gencode. Such case are treated the same as missing entries, described above. An alternative would be to provide a comma-separated list of mapping targets and their chromosome offsets and or ranges. As this situation tends to only occur in older UCSC reference genomes, which are decreasingly used, I would prefer to avoid this complication.
It's often the case that a patch will have an associated version, such as
KB469738.2. While the patch itself will exist across genome updates, the version number may change. Consequently, it may be required to strip off these version when performing name conversions, simply to support different versions/patches of the same genome.
Note that some data sources are absent. For example, wormbase has not been included, since it's chromosome naming system is identical to that in Ensembl.
Please submit a pull request or an issue if you find any errors!