I am using UCSC's liftover tool to port a number of genomic coordinates between hg19 and hg38. A tiny fraction of coordinates from hg19 does not map to hg38. This is not unexpected, however, when I try to convert these coordinates using the Ensembl Assembly converter, some of the coordinates actually map to a position in hg38.

Now I'm a bit confused o_0 - does anyone know what difference in the implementation might cause this? Are there any caveats associated with either method?

Listed below some of the coordinates I can convert with Ensembl Assembly converter but not UCSC LiftOver (using the respective web-interfaces). Maybe I am making a mistake with the 0-/1-based systems in UCSC/Ensembl but I can't see it...

• LiftOver: chr19:42366232-42366233 -> no map
• Ens: chr19 42366233 42366233 -> 19:41862160
• LiftOver: chr1:145442254-145442255 -> no map
• Ens: chr1 145442254 145442254 ->1:145990066