I'm looking to convert a list of SNP coordinates from hg18 to hg19 using liftover, but I'm worried about the resolution of LiftOver. If LiftOver gets a SNP position wrong by even a single base, then that SNP becomes useless. So is LiftOver accurate and precise enough to get most of my SNP positions exactly right in the new genome?
Alternatively, if anyone knows where to get huref variants in hg19 coordinates, that works too.
You should understand how liftOver lifts Over. Let's assume you will use liftOver to lift from hg18 to hg19. Someone has blasted regions of hg18 against hg19 and created a lookup table (map) for all positions that blasted perfectly to hg19. When you use liftover you basically skip this time consuming blasting step and use the already prepared map to determine which base in the hg19 correspond to your hg18 base. Per this definition and method, if you have a corresponding baseposition in hg19 (read your target) you should have enough confidence in your liftOver process to provide you with the correct target position. That being said, like every good paranoid scientist and bioinformatician(I included) you will probably check a handful of randomly selected liftedOver positions for known snps etc and make sure for your self that it is indeed the case.
UCSC's liftover is accurate and precise enough for what you are asking. in fact, if it wouldn't be as accurate as stated, its existence just wouldn't make sense. what it can happen is that some of your positions may not be translated, but the output report will warn you about those issues (if any).
regarding the retrieval of huref variants in hg19 coordinates... what do you mean with huref variants? is it human variants mapped to the current GCHr37 reference? would ensembl's biomart meet your requirements, since there are several "general variation filters" you can use for your query?
Yes, liftover does.