Bwa Alignment Length

Question

[Newvin]

After performing a transcriptome assembly, I am using BWA to align the reads to my assembled contigs. From the resulting SAM file, I am able to calculate the # of reads that map to each contig, but I would like to normalize this abundance measure by using the length of each alignment.

My naive question: is the length of each alignment equal to the length of the read that is being aligned, or does BWA sometimes align only part of a sequence query to the target?

[Damian Kao]

What sequencing platform are you using? Your read lengths vary that much?

[Newvin]

These are Illumina short reads. The variation in read length occurs because we used an adaptive algorithm to trim the reads based on quality score prior to assembly.

Answer 1

[Jeremy Leipzig]

BWA is performing a local alignment. If you see an 'S' in the CIGAR string of the resulting SAM file (e.g 40M36S), that 3' soft clipped end is not part of the alignment.