Hi Mike, I went to the scientist that created the TransMap track for advice, here are his comments:
Thanks, Jennifer > > 1. Of the 26 listed prediction programs, what prediction program do > > you recommend for this? Modern computational gene finder are complex to run and require a fair amount of experience to set the parameters. One is much better off using From my experience Ensembl, and N-SCAN on Dog/Cat and Augustus on cat have excellent reputations and would all be excellent choices. > > 2. Would transmap work? TransMap provides good accuracy, since it's based on orthologous transcripts in other species and is very good and show the details of gene evolution. It will not show novel genes or isoforms and is limited by the transcripts in the source species. > > 3. Transmap only lists a couple species on the full option. Why? > > Does this exist for dog or cat? TransMap uses the existing cross species nets and chains and existing source transcripts. For mammals, only human and mouse have excellent cDNA coverage. TransMapped transcripts exist for from other species, but there are far fewer of them. Additionally, cat will be limited by the smaller number aligned species. However human and mouse should provide good coverage. > > 4. Would you stick with only one program or multiple program for > > this? All methods have their strengths and weaknesses. Our experience with the MGC project has show that predictions where multiple sources agree are much more likely to be real genes. ------------------------------------------------ Jennifer Jackson UCSC Genome Bioinformatics Group _______________________________________________ Genome maillist - [email protected] https://lists.soe.ucsc.edu/mailman/listinfo/genome
