Differences in results of Bowtie2

Question

[snehu.sambare]

Bowtie 2 yields different results when "-a" option is used.

bowtie2 **-a** -x ../../masked_rDNA ../hTERT-TP2-R1_p1.fq --no-unal > temp.sam
16914184 reads; of these:
  16914184 (100.00%) were unpaired; of these:
    16775370 (99.18%) aligned 0 times
    **98594** (0.58%) aligned exactly 1 time
    **40220** (0.24%) aligned >1 times
0.82% overall alignment rate

bowtie2  -x ../../masked_rDNA ../hTERT-TP2-R1_p1.fq --no-unal > temp.sam
16914184 reads; of these:
  16914184 (100.00%) were unpaired; of these:
    16775461 (99.18%) aligned 0 times
    **98556** (0.58%) aligned exactly 1 time
    **40167** (0.24%) aligned >1 times
0.82% overall alignment rate

Since my analysis is sensitive to the counts, I needed to know the reason for such differences.

Answer 1

[GenoMax]

As you can see you have really low alignments with <1% of your reads mapping.

This may be a good result though since your file names have rDNA so perhaps you are looking for rRNA contamination and its absence may be what you are interested in. -a reports all multi-mapping reads but it does not look like you have many that fall in that category.

[snehu.sambare]

I am interested in extracting rDNA sequences from the fastq file. If I run Bowtie using -a option, I am getting (98594+40220) 138814 reads as rDNA, but without -a in the second run, the reads I get are (98556+40167) 138723 reads are rDNA.

So the second run has 118 reads less than the first run.

My concern is I may lose something important if I don't use -a option.

So I was wondering what changes, when I use the -a option?

[GenoMax]

With -a you are getting all multi-mapping read locations. I don't recall if bowtie2 does not report reads that align over a certain limit (check what is default value for -k) but specifying -a would report all of them.

[snehu.sambare]

Thank you. This was helpful :)