Hi Jean-Francois, One of our developers had the following comments:
" It's very likely that some of the calls would change if they re-ran their calling on hg19, versus what they'd get if they lifted over from hg18, but most would not change. There are a few regions in hg19 that are very different than what is in hg18. Most of these regions can be identified by looking at the "hg18 Diff" track in the Mapping and Sequencing Tracks group on hg19, but there may be regions in hg19 that aren't in hg18 at all, that wouldn't show up in the Hg18 Diff track. There are also mapability issues that would differ slightly between hg19 and hg18, so where reads mapped could vary in subtle ways that might affect the variant calling. So, long story short, liftOver would give them answers quickly, and most of the calls would probably end up in the same place if the genotyping was redone, but there will probably be differences as well, especially in regions that have dramatically changed between hg18 and hg19 (which can be identified in the *Diff tracks)." Please let us know if you have any additional questions: [email protected] - Greg Roe UCSC Genome Bioinformatics Group On 10/17/11 10:08 AM, Jean-François Payotte wrote: > Hi, > > This may be a somewhat obvious question, but I can't seem to find a clear > answer anywhere. > > We are currently working with genotype data (where genotype-calls were made > using Affymetrix birdseed software) annotated with human assembly hg18 (NCBI > 36). Now, we consider switching our data to human assembly hg19 and I was > wondering if we would need to redo the genotype-calling procedure using > directly hg19 annotations? > > I am aware of the UCSC "LiftOver" software, which allows to convert data > between different human genome assemblies, but I would like to know if this > method would be sufficient or if it would be a better practice to redo the > genotype-calling for our data. > > Many thanks, > Jean-Francois > _______________________________________________ > Genome maillist - [email protected] > https://lists.soe.ucsc.edu/mailman/listinfo/genome _______________________________________________ Genome maillist - [email protected] https://lists.soe.ucsc.edu/mailman/listinfo/genome
