From genomewiki
Jump to navigationJump to search

HgLiftOver is a Genome Browser utility for converting genome coordinates between assemblies.

This page could be expanded to include all kinds of information about hgLiftOver. For now it just has a section on a point of confusion about hgLiftOver and BED format.

hgLiftOver - BED vs. Position format

Sometimes when one converts coordinates using hgLiftOver using position format and then tries it again using BED format, it looks like the BED format results are incorrect. This happens when (1) the input used for BED is the exact same as that used for position (i.e., the first coordinate is not lowered by one to turn it into BED format) and (2) the converted section in the new assembly is on a strand with the opposite orientation of the original section (i.e., the net and chain tracks appear to be on the negative strand).

Here is an example of this potentially confusing case -- conversion of the rn4 gene Apo-E to hg18 coordinates.

First, the hgConvert results from rn4 to hg18:

 Rat Nov. 2004 chr1:79003641-79006387 to Human Mar. 2006
 chr19:50100923-50104476 (73.9% of bases, 100.0% of span)

And the exact same results from hgLiftOver, using the POS option:

 chr1:79003641-79006387 (rn4)
 chr19:50100923-50104476 (hg18) <-- same as hgConvert.

Now, if we try to convert the exact same coordinates using the BED option and just take out the position punctuation so that it looks like BED format (this is the incorrect way to do it):

 chr1 79003641 79006387 x (rn4)
 chr19 50100922 50104475 x 1 (hg18) <-- in Browser position coordinates, this is chr19:50100923-50104475.

This is NOT the same result as we got with position.

Here is a BED conversion using the correct coordinates for BED format (the first position format coordinate is lowered by one):

 chr1 79003640 79006387 x (rn4)
 chr19 50100922 50104476 x 1 (hg18) <-- in Browser position coordinates, this is chr19:50100923-50104476.

The same result as the position method!