GrCh37 or GrCh38? That is the question.

Question

[Anjan]

I'm getting back into human genome analysis after about 10 years. In 2015, GrCh38 was still considered to be in "beta version"- with several misassemblies and misannotations. GrCh37 was the recommended version. Have things changed since then? Which version would you recommend today? Thank you.

Answer 1

[GenoMax]

GRCh38 is what you should use, as it will be best supported by most tools (in terms of annotation etc). Heng Li has a blog post https://lh3.github.io/2017/11/13/which-human-reference-genome-to-use

If you want the most complete assembly currently known then, the T2T genome would be the one to use: https://www.ncbi.nlm.nih.gov/datasets/genome/GCF_009914755.1/

[Anjan]

Thank you @genomax. Heng Li's blog was helpful.

Answer 2

[Papyrus]

Using GRCh37 nowadays is out of the question, and while I'd say GRCh38 is probably still the standard used in many genomic pipelines, you should definitely look up the T2T (telomere-to-telomere) reference, published in 2022, which will end up becoming the new standard for many uses. Also check out this nice Biostars thread comparing hg38 and T2T.

Answer 3

[Pierre Lindenbaum]

Grch37 is old. Do not use T2T if you need any common database like gnomad.