Unmapped paired end reads with confusing sam flag values

Question

[irritable_phd_syndrome]

I have two reads (paired end) in my bam file generated by bowtie2.

K00194:35:H7JHYBBXX:1:1101:10653:1349   77      *       0       0       *       *       0       0    ACTGAATT  JJJJJJJJ  YT:Z:UP

K00194:35:H7JHYBBXX:1:1101:10653:1349   141     *       0       0       *       *       0       0       GATTCGCC  JJJJJJJJ YT:Z:UP

I am confused about the flag values. The first flag score (77) makes sense (0x1 = 1 = multiple segments, 0x4 = 4 = unmapped, 0x8 = 8 = next segment unmapped, 0x40 = 64 = first segment in template. The next score (141) I find confusing (0x1 = 1= multiple segments, 0x4 = 4 = unmapped, 0x8 = 8 = next segment unmapped, 0x80 = 128 = last segment in template). The confusion is the last two bits (0x8 and 0x80).

How can you have the next segment unmapped when you are the _last_ segment in the template? I would think that the value should be 133 instead?

Answer 1

[John]

The "next" value for the last read/segment always wraps around in the SAM format. It's true also for flags 0x8 and 0x20, as well as RNEXT and PNEXT.

PNEXT: Position of the primary alignment of the NEXT read in the template. Set as 0 when the information is unavailable. This field equals POS at the primary line of the next read.

Answer 2

[swbarnes2]

The flags say that the reads are paired, and that both are unmapped, and that one is read 1, and one is read 2. The "8" column refers to the mate, not the "next" segment.

https://broadinstitute.github.io/picard/explain-flags.html

[John]

That is how Picard defines flags and not strictly the SAM spec.