Different position of the same alternative locus

Question

[ivar.grytten]

Where do people get the position of GRCh38 alternative loci?

I've found multiple sources, and they seem to give different positions of some alternative loci.

If I download the patch 9 dataset from http://www.ebi.ac.uk/ena/data/view/GCA_000001405.24 (by clicking "regions" in the top right corner), I will for get that the alternative locus KI270915 is positioned on chr 19:54698377-54869056:

LRC_KIR KI270915.1 170665 1-170665 19 54698377-54869056

However, If I download patch 9 from ncbi (ftp://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/000/001/405/GCA_000001405.24_GRCh38.p9), the same alternative locus is positioned on chr 19:54025634-55084318

LRC_KIR 19 54025634 55084318 alt-scaffold KI270915.1 NT_187669.1 ALT_REF_LOCI_21

Please correct me if I have misunderstood the format of these files.

I have not checked all the alternative loci, but most of them seem to have identical coordinates between the two sources, apart from a few.

Does anyone know why these would be different?

Also, does anyone know how the position of these alternative loci are decided (see this unanswered question: How are the GRC alt locus to primary assembly alignments generated? )?

---

Answer 1

[deanna.church]

What NCBI produces is exactly what the GRC submit. Many other sources will re-align the alt-loci to the Primary assembly. We don't really treat alignments as a first class data object (no accession, versions, etc) so people feel free to do this. I think it just leads to confusion (as noted in the question).

That is not to say there is not room for improvement in the alignment process- but many of these alignments are in really tough regions. The alignments are generated by a tool at NCBI called NGAlign. You can find this in the Genome Workbench tool distributed by NCBI as well. Some of the alignments end up getting manually curated to adjust particularly challenging regions.

[ivar.grytten]

Thanks, that answers my question