Bowtie2 Alignment summary

Question

[bassanio]

Given below is a summary of Bowtie2 alignment result of mapping transcriptome data to REF genes.

Command used:

bowtie2 -q -N 1 -p 8 -x REF/REFgene.fasta -1 DATA1/S1_1.fastq -2 DATA1/S1_2.fastq --al-conc Sample_unmapped.fastq -S REF/Bowtie_Result/REF/S1.sam

Alignment summary:
474285 (100.00%) were paired; of these:
    413631 (87.21%) aligned concordantly 0 times
    58324 (12.30%) aligned concordantly exactly 1 time
    2330 (0.49%) aligned concordantly >1 times
    ----
    413631 pairs aligned concordantly 0 times; of these:
      251838 (60.88%) aligned discordantly 1 time
    ----
    161793 pairs aligned 0 times concordantly or discordantly; of these:
      323586 mates make up the pairs; of these:
        291974 (90.23%) aligned 0 times
        8787 (2.72%) aligned exactly 1 time
        22825 (7.05%) aligned >1 times
69.22% overall alignment rate

My few concerns:

1. 87.21% aligned concordantly but my overall alignment decreased to~ 70%. why does it comes down?
2. I tried to fetch the unaligned reads using --un-conc but it gives me concordantly aligned reads count but the concordantly unaligned results is obtained through -al-conc

Answer 1

[thackl]

87.21% aligned concordantly 0 times - meaning they didn't align concordantly..

[bassanio]

Hi,

Thank you, I didn't notice that. What I would like to do is to fetch the unaligned reads. if I use --un-conc I get 413631 reads but its like 87% of total reads by my overall alignment rate is 69.22%. if I use --un I don't get any results. So how to fetch those reads(PE) that are not mapped.

[thackl]

You want to have pairs with both reads not aligned? I'm not entirely sure, but I think you could use samtools view with an appropriate filter to get the alignments you want and than use samtools bam2fq to get the reads. Something like this maybe:

bowtie2 ... | samtools view -f 13 ... > unmapped-pairs.bam
samtools bam2fq unmapped-pairs.bam > unmapped-pairs.fq

Answer 2

[Devon Ryan]

~13% aligned concordantly, ~54% aligned discordantly (did you swap read 1 and 2 or is the insert size just non-standard?) and another percent or two aligned as singletons. That's where the ~70% comes from.

[bassanio]

Hi,

Thank you insert is of varying length. I understood how you came up with 13% of concordantly : 12.30%+ 0.49%. But how did you came up with 54%? Secondly my aim is get the ~30% reads that didn't aligned.

[Devon Ryan]

60% of 87% is ~54% (it's actually 52.2%, but I was doing this in my head). ~87% of your reads weren't aligned concordantly, of those 60% were then aligned discordantly. Therefore, ~52% (I should have just used a calculator the first time) of your reads aligned discordantly.

If you want to get the reads that don't align the simplest method is to just use samtools to extract them from bowtie2's output (see the most recent comment from https://www.biostars.org/u/14801/).