Bowtie2 unpaired mode usage

Question

[Jimpix]

Hey! Taking into consideration commands below:

bowtie2 -x Bowtie2Index/hg19 -U R1.fastq, R2.fastq -S bowtie2_unpaired.sam 

and

bowtie2 -x Bowtie2Index/hg19 -U R1.fastq -S bowtie2_unpaired_R1.sam 
bowtie2 -x Bowtie2Index/hg19 -U R2.fastq -S bowtie2_unpaired_R2.sam 

There is any difference between files bowtie2_unpaired.sam and merged by pysam files bowtie2_unpaired_R1.sam and bowtie2_unpaired_R2.sam? Thanks for the answer.

[GenoMax]

Bowtie 2's search for alignments for a given read is "randomized." That is, when Bowtie 2 encounters a set of equally-good choices, it uses a pseudo-random number to choose. For example, if Bowtie 2 discovers a set of 3 equally-good alignments and wants to decide which to report, it picks a pseudo-random integer 0, 1 or 2 and reports the corresponding alignment. Arbitrary choices can crop up at various points during alignment.

Not sure if you set the same --seed for above commands you would get the same alignment for each read in the two operations.

[Istvan Albert]

basically what this says is that one would not necessarily get identical BAM files even if they ran Bowtie on the same file.

[GenoMax]

if --seed is set the same for both runs, Bowtie 2 will produce the same output; i.e., it will align the read to the same place, even if there are multiple equally good alignments

If one sets the same --seed then I was wondering if the results from the two options above will be identical for individual reads. With --non-deterministic option bowtie2 will report different alignments.

Answer 1

[Istvan Albert]

Merge operates on sorted BAM files. All inputs should be sorted and the outcome is another sorted BAM file.

What you show above is a concatenation operation, the BAM files will follow sequentially.

So in general you would not expect the two operations to produce identical order of the alignments.