How Does Ucsc Liftover Handle Loci That Map To Multiple Genome Positions?

Question

[Andrea_Bio]

Hi

I'm just looking at liftover to map some SNPs from one genome assembly to another and the output i get is a file of mapped snps and a file of unmapped snps. The unmapped snps have a short statement describing cause of failure e.g. deleted in new

1) Is there any documentation explaining all of the causes of mapping failure in liftover as i didn't get any when i downloaded the executable

2) How does liftover handle positions that map to multiple genome locations. I don't know whether they class as failures or whether you would get multiple rows in the 'mapped snps' file with one row per mapped position

thanks

Answer 1

[Pierre Lindenbaum]

1) I don't think you can find a verbose 'reason' to explain with liftOver failed. In the case of "Deleted in New" this would be just None of sequence intersects with any alignment chain for the region .This message can be find in the source code: http://hgwdev.cse.ucsc.edu/~kent/src/unzipped/hg/lib/liftOver.c

2) as far as I can see in the code, liftOver " sorts (the) chains by position in target to order subregions by orthology" and then displays each "hit chain"

[Andrea_Bio]

i haven't programmed in c for years! I dont really know what the answer for (2) means. Does that mean it classes inputs that map to multiple locations as pass or fail?

Answer 2

[apa@stowers]

Six years later there is still no web documentation for #1, but you might look into the "-errorHelp" flag.

For #2, liftOver will only return one->one mappings by default, and will discard any coords mapping to multiple locations. If you want to allow one->many mappings, use the "-multiple" flag.