SourceForge has been redesigned. Learn more.

Paired-end mapping with -a all reporting

  • Craig Jackson

    Craig Jackson - 2012-10-09


    I have a question about how Bowtie2 will report paired-end alignment with -a all alignments reporting.

    From the Bowtie2 poster, it seems Bowtie2 is set up to find discordant pair alignments only if a concordant pair alignment can not be found, and then only find unpaired alignments if no discordant pair alignments are found (Figure 3). It also seems this way from the way the alignment results are reported (

    Does this change when using the -a reporting mode? Will discordant and unpaired alignments that pass the -min-score threshold still be reported even when a concordant alignment is found? Or will the reporting still follow the 'if concordant else if discordant else unpaired' model, and just report all/multiple concordant alignments and skip the discordant and unpaired alignments since concordant alignment was found.

    I am using -a mode because the reference genome is highly duplicate gene/region-filled, and I want the mapping to report the read alignments to all of the duplicates, not just the best, in order to do a read depth copy number variation study. For this case, would it be preferable to tell Bowtie2 to map the paired-end reads as just single reads (use -U instead of -1 and -2)?



Log in to post a comment.