On Sat, Feb 26, 2011 at 5:56 PM, Ravi Karra <ravi.karra@gmail.com> wrote: > Hi Sean, > I could do that, but am not sure how to. The annotated zebrafish genome is > the Tubingen strain and the some of the probes on the array are from the EK > and AB strains. This means that I need to allow for SNP's in the > alignment. I originally tried to align the probes to the Ensembl > Transcripts using matchPDict but when I allowed for 2 mismatches > (max.mismatch = 2) across the probe sequences, my computer never stopped > running the program! I found TopHat to be much faster (8 min) and TopHat > allows for a few nucleotide wobble by default!. > Do you have a suggestion(s) on another way to align the array to the > Ensembl Transcripts? > Hi, Ravi. You could try blat, blast, gmap, or ssaha, for example. I have used blat and gmap successfully to annotate probes. Against transcripts, blat or gmap should run in seconds to minutes. Sean > Thanks, > Ravi > > > > > On Feb 26, 2011, at 5:45 PM, Sean Davis wrote: > > > > On Sat, Feb 26, 2011 at 5:29 PM, Ravi Karra <ravi.karra@gmail.com> wrote: > >> Hello, >> >> I am still trying to map probes on the Nimblegen Zebrafish 12 x135K >> Expression to the Zv9 version of the zebrafish genome available from >> Ensembl! I am very reluctantly pursuing an alignment approach to annotation >> as the original annotation provided with the array is quite outdated. >> >> I performed a gapped alignment using the individual probe sequences >> (60-mers) from the array using TopHat and loaded the results into >> Bioconductor as a GappedAlignments object. I have made a TranscriptDb >> object using the Zv9 genome from Ensembl. Next, I plan to use findOverlaps >> for the annotation. What is the best way to get the overlaps (by exon, cds, >> or by transcript)? I am a little concerned that using transcriptsByOverlaps >> might be a bit too broad and result in mapping reads to multiple genes (for >> example transcripts in the genome that have overlapping genomic >> coordinates). By contrast, mapping with exonsByOverlaps and cdsByOverlaps >> might be too restrictive and miss information in the UTR's. My gut feeling >> is that annotating by cds is the "in-between" approach. What is the >> recommended approach for RNA-seq? As you can tell, I am quite new to this! >> >> > Hi, Ravi. Sorry to answer a question with more questions, but why not just > map the probes against Ensembl Transcripts or refseq? What is the advantage > of mapping to the genome and then going back to the transcripts? > > Sean > > > [[alternative HTML version deleted]]