Do coordinates change between minor ref genome builds and can Liftover be used to convert minor ref genome coordinates?

Question

[[e-mail]]

Two questions:

1) Do coordinates ever change for minor reference genome builds compared to the major reference genome build?

• For example, are coordinates for coordinates aligned to GRCh38 major reference genome always going to be the same as coordinates aligned for all minor reference genomes of GRCh38 such as GRCh38.p2, GRCh38.p3, GRCh38.p4, etc?

2) Can Liftover and other tools for converting coordinates between builds (such as listed here: https://www.biostars.org/p/65558/) be used to convert coordinates from a reference genome minor build to coordinates for a different major reference genome?

• For example, can Liftover and all those other tools be used to convert coordinates aligned to GRCh38.p3 to GRCh37 coordinates?

Answer 1

[Devon Ryan]

1. No, the difference is only in "patch" contigs and alternate loci, which end up getting added. I highly recommend having a read through this for further information and confirmation of this.
2. As long as you're not trying to convert coordinates on a patch contig or alternate locus then everything should work.

Answer 2

[Denise CS]

Just a few extra pieces of information in addition to Devon's:

1) check the summary statistics tables for three different patches e.g. GRCh38p3, GRCh38p2 and GRCh38p5. You will see the Golden Path Length is the same, i.e. 3,096,649,726. That's the primary assembly, without patches and haplotypes (i.e. alternative sequence).

Minor releases (adding patches/haplotypes to GRCh38 or GRCh37) do not lead to changes in the coordinates of the primary assembly. Although different assemblies are necessary (and more accurate than the previous counterpart) to fill gaps, correct sequencing errors, etc, changes in the assembly are not welcomed by people working on genomics. So the minor releases are the compromise found by the Genome Reference Consortium.

2) Assembly converter tools are designed to convert coordinates from GRCh38 to GRCh37 or NCBI36 (and vice-versa), not from minor releases, for example GRCh38p3 to GRCh38p2.