Hello, Yasmina.
Thank you for your interest in the Genome Browser.
We would need to evaluate the oddities on a case by case basis, but the location of the variant you provided on hg38 (17:138824) is located on a completely new sequence that was added to hg38 and does not exist on hg19 (
https://www.ncbi.nlm.nih.gov/nuccore/AC240565.4).
You can see the following session which shows the entire new contig in place (
http://genome.ucsc.edu/s/Lou/RM30088). Looking at the self-chain track, which is what liftOver is built on, it looks like there is a fairly high quality sequence match between this new region and the location on hg19 chr11, which is why liftOver ultimately matches it there. That being said, it is an entirely novel region on hg38. Sometimes the GRCh moves a contig to a different location in assembly updates (even to different chromosomes), however when that happens it is annotated as such. That is not the case here, so it would appear there is no exact corresponding location of this variant on hg19.
If your variant has a dbSNP rsID it is always recommended you attempt to pick up the corresponding coordinates on another assembly using the ID and not liftOver (
http://genome.ucsc.edu/FAQ/FAQreleases.html#snpConversion). That being said, after spot checking some variants located in this new contig I don't believe any have corresponding hg19 coordinates, likely due to it being novel sequence.
I hope this is helpful. Please include
gen...@soe.ucsc.edu in any replies to ensure visibility by the team. All messages sent to that address are archived on our public forum. If your question includes sensitive information, you may send it instead to
genom...@soe.ucsc.edu.
Lou Nassar
UCSC Genomics Institute