There are multiple versions of human reference genomes. The most widely used so far is GRCh38 or hg38 according to UCSC. NCBI, GRC, UCSC and Ensembl all provide their versions of GRCh38, which are all subtly different from each other and are often suboptimal. It is surprisingly nontrivial to choose the right reference genome in your analysis.
This repo provides the URLs of several human reference genomes I recommend. These genomes were adapted from the original GenBank verions and work better for read alignment and variant calling. All of them include unlocalized and unplaced contigs, use rCRS for mitochondrion and have pseudoautosomal regions (PARs) on Y chromosome hard masked.
To download these genomes, see links in URL.txt. The genomes are also "permanently" available at doi:10.5281/zenodo.8045373 along with their BWA and Bowtie2 indices. Command lines for generating secondary files can be found in ref-gen.mak. Note that downloading from Zenodo can be slow. I will look for a faster host in future.
The following table gives a brief description of these genomes:
| Name | Coordinate | Brief description |
|---|---|---|
| hs37 | GRCh37 | 1000 Genomes Project (1000G) reference in 2010 |
| hs37d5 | GRCh37 | hs37 with decoy (recommended GRCh37) |
| hs38 | GRCh38 | GRCh38 no-alt analysis set (recommended GRCh38) |
| hs38DH | GRCh38 | GRCh38 with ALT, decoy and HLA genes (not recommended) |
| chm13v2 | CHM13v2 | T2T CHM13 plus HG002 chrY (recommended CHM13) |
In more detail:
-
hs37. This was the reference genome used by the 1000 Genomes Project at the early stage. It was derived from the Ensembl primary genome but had mitochondrial DNA replaced with rCRS.
-
hs37d5. This was generated on top of hs37 and used by 1000G in 2011. The only difference is the addition of decoy sequences and the EBV genome. This version has been found to give better variant calls. Recommended version if you use GRCh37.
-
hs38. This is the so-called GRCh38 no-ALT analysis set. Recommended if you use GRCh38.
-
hs38DH. This genome includes GRCh38 alternate contigs, GRCh38 decoy sequences and HLA alleles. As GRCh38 is more complete than GRCh37, GRCh38 decoy sequences are not as important as GRCh37 decoy. Furthermore, most tools would not work well with this version as they are not ALT-aware. Improper use of this genome would hurt variant sensitivity. Not recommended unless you know what you are doing. hs38DH has been used to create high-coverage 1000G CRAM files. You may need this verion to decode these CRAM files.
I was involved in discussions on generating hs37 and hs38. I created the final FASTA files of hs37d5, hs38DH and chm13v2. Of course, the original genome sequences were produced by others. If you are confused by hs38DH, you may blame me for that.
Note that the description above is consistent with my old blog post on "Which human reference genome to use?" This time I added chm13v2 as well as download links to packaged sequences and indices.