BWA vs Bowtie2 output

Question

[kimaebeloe]

Hi,

I'm new in this business, but I hope my question make sense :)

I'm trying to compare the output from BWA and BT2 from the same mapping.

As I understood, all matches in the sam-files are represented by each line not in the header (i.e. those that do not start with '@')

In the BWA sam-file there are many more of those lines compared to the BT2 sam-file. Moreover, I see that many of the matches in the BWA sam-file have code 4 in the FLAG-column. If I understand the sam-documentation correct, FLAG 4 means that the read is not a match. So how do I interpret a 'match' in the BWA sam-file with FLAG 4?

Again, I am new in this branch, but hope the question makes sense.

Take care!

Kim

[Zaag]

Could you show both lines?

[kimaebeloe]

How do you mean?

There are a lots of lines. But in the BT2 sam-file none of them has the value 4 inte the FLAG column whereas many output lines in the BWA does.

[iraun]

According to Devon's answer, you could try to remove unmapped reads and secondary alignments from bwa output bam file using samtools, and then compare both files.

[kimaebeloe]

OK - thanks again!

Just out of curiosity, if unmapped segments ((FLAG=4) is segment the same thing as a read?)) makes it to the alignment section of the SAM-file, which ones, then, does not make it to the alignment section?

[Devon Ryan]

Yes, if a read is unmapped it'll be a single segment. BTW, bowtie2 won't produce alignments of reads in multiple segments (though bwa mem will).

Answer 1

[Devon Ryan]

BWA will output unmapped and secondary alignments, which bowtie2 won't do by default. If bit 4 in the FLAG field is set, then you should ignore any coordinates or other related information for an entry.