Hi genome gurus, there is something I've always wondered about in your multiple alignments, it's how the mafs are produced from the chains. As I've only used pairwise alignments until now, I was fine with netToAxt and never really understood what multiz was doing.
My impression of the whole alignment procedure was that fragments are aligned using lastz, alignments are combined into chains and nets and the nets are used to pull out again the best chain-alignments. Then multiz/autoMZ is reformatting the filtered chains into maf alignments. So I thought the the chain alignments should correspond to the final maf alignments. However, I have the impression that multiz is also doing some realignment of the chains and is not only reformatting them. People have asked me several times where these differences come from that are shown on the browser, eg. if I go to chr21:33,035,903-33,035,993 on hg19, dog and mouse deletions more or less correspond between chains and mafs, but they are also often shifted by some basepairs, introduce new gaps or split gaps into several pieces. Here is a marked up screenshot of this region with the mouse & dog differences highlighted: http://markup.io/v/zp4d8daatpnm Can you give me a hint what I am missing ...? cheers Max PS: Thanks for the amazing right-click navigation! It just got so much easier to explain the website to beginners!! _______________________________________________ Genome maillist - [email protected] https://lists.soe.ucsc.edu/mailman/listinfo/genome
