From about 1 million segments from hg19 of the form:
chr2 111032430 111052520
I tried to re-map these to hg18 using LiftOver. LiftOver performs very well and was able to correctly map all but about 1,500 reads to the old hg18 release. In the interest of being thorough, I wanted to find an automated way to map the remaining releases back to hg18 if at all possible. So, for instance, the above (according to BLAT) would re-map to
2 111169575 111170890
2 110191298 110210043
And so on for each. Right now my data is in Galaxy. Is there a simple tool to re-map any un-mapped segments from hg19 to hg18 where LiftOver fails?
Thanks,
Rx
•
written
by
Ryan D
1 answer
Right now my data is in Galaxy. Is there a simple tool to re-map any un-mapped segments from hg19 to hg18 where LiftOver fails?
To answer your question: there is no simpler tool than LiftOver. As Brent mentioned try to tune the parameters. The next step is what you propose: align each segment with a method of your choice.
Log in to answer this question.
did you try messing with the parameters to liftOver, e.g. -multiple and lowering -minMatch?
That's cool, Brent. That got me most of the way there. Changing "minimum number of bases that must remap" from 0.95 to 0.50 and "Multiple output regions permitted" to No. That left me with 135 regions un-mapped. I tried to re-run those remaining 135 thinking they may be split. Unfortunately LiftOver crashes (in Galaxy) with this :( But I'm almost there. Thanks.