Bowtie2

Link to Bowtie2 manual.

Bowtie2 is the aligner of choice for our application (eukaryotic small genomes) according to our collaborators (JB and DD). However, we are now considering using BWA_MEM because, in our hands, it seems to produce cleaner alignments and consequently is more conservative about SNP calling. Here is a collection of thoughts on how Bowtie2 should be implimented in our context:

• It turns out SAM/BAM headers are very important for GATK. Specifically, the @RG line is crucial, and is not included in the SAM header by default. Bowtie2 should be tole explicitly to include the most important components of the @RG line: RGID, RGPL, RGLB, RGSM. Adding the header at this step means we don't have to add important header information later, using Picard's AddOrReplaceReadGroups, for example.
• If you view a SAM file using cat foo.sam | more, the header is obvious. However, if you view a BAM file using samtools view foo.bam | more, the header isn't there. That doesn't mean that the header isn't there. You can view the header by specifying samtools view -h foo.bam | more.