Question about samtools view flags (paired reads vs. properly paired reads)

Question

[mrk]

What exactly does it mean for a read to be 'mapped in a proper pair' (flag 0x2) as opposed just paired reads (0x1)? I have a bam file where, when I use -f 0x2, I get no output. Does this mean the reads weren't aligned or something?

Answer 1

[Pierre Lindenbaum]

paired reads = the experiment was a paired end assay: each fragment was sequenced on 5' and 3', two FASTQ files are generated by the sequencer.

mapped in a proper pair= the two sam records of a pair are mapped on the sam chromosome, at a good distance, mapped on forward and reverse strand (like a PCR, but it's not a PCR)

[mrk]

Thanks. What determines what constitutes a good distance? I have some targeted panel data and looking at flagstat, almost all reads are paired, mapped, and 'with itself and mate mapped' but 0 are properly paired. The distance might be the issue?

[Pierre Lindenbaum]

Thanks. What determines what constitutes a good distance?

I think tools like BWA use the median distance for each chunk of processed data