Hello Peng, Please see our comments with your questions below.
Thanks! Jennifer ------------------------------------------------ Jennifer Jackson UCSC Genome Bioinformatics Group ----- "Jennifer Jackson" <[email protected]> wrote: > From: "Jennifer Jackson" <[email protected]> > To: "genome" <[email protected]> > Cc: "Peng Yu" <[email protected]> > Sent: Wednesday, January 6, 2010 4:24:21 PM GMT -08:00 US/Canada Pacific > Subject: Re: [Genome] algorithm in liftOver > > Repost: > > Subject: Re: algorithm in liftOver > Sent By > "Peng Yu" <[email protected]> On: January 5, 2010 12:55 PM > To: [email protected] > > On Tue, Dec 22, 2009 at 3:47 PM, Peng Yu <[email protected]> wrote: > > I don't find a paper that describes how liftOver works. Could > somebody > > let me know the algorithm used in liftOver? The liftOver program using the Chain track information. Please review the track documentation for the Chain track processing to understand how this is done. The usage for liftOver can be found by typing in the program at a unix prompt. Pre-compiled liftOver executables are available via ftp from here = http://hgdownload.cse.ucsc.edu/admin/exe/ The input over.chain files are located in the "from" species downloads location under the header "LiftOver" In general, this is the process for Chain -> Over.chain liftOver file: a) BLAT is used for same-species liftover and blastz for everything else. b) To create chain files describing how blocks from two species align: + in chain file means that + in orgA maps to + in orgB, and vice versa - maps to - - in chain file means that + in orgA maps to - in orgB, and vice versa - maps to + > > Are the chain files be generated by aligning two version of gnomic > > sequence (say mm6, mm8)? > > > > Why there are no direct map between, say, mm6 and mm8? Yes, two complete genomes are compared. Not all pairs have liftOver chain.over files, but there is almost always a path from one release to another (if not direct, then through other assemblies). We have determined that for incremental releases of the same genome, this does not affect the liftOver results. For cross-species lifts, you might need to experiment to find the best parameters to maximise the quality and/or volume of the results. > > > > liftOver accept BED files? But the strand field is optional in BED > > files. Does liftOver guarantee that a gnomic location has to be map > to > > the same strand ('+' or '-'), if the strand field in BED is not > > specified? > > > If the BED file to be lifted has no strand specified, + will be assumed. > > > ------------------------------------------------ > Jennifer Jackson > UCSC Genome Bioinformatics Group > _______________________________________________ > Genome maillist - [email protected] > https://lists.soe.ucsc.edu/mailman/listinfo/genome _______________________________________________ Genome maillist - [email protected] https://lists.soe.ucsc.edu/mailman/listinfo/genome
