Inconsistent pairing in SAM file?

[syfo]

In that SAM file, read1's mate is read2 but read2's mate is not read1. Is that possible? Both are on chrX, convergent orientation (-> <-), perfectly mapped (36M).

HWI-M00185:3:000000000-A1BH4:1:1101:5903:5381   353     chrX    66951034        254     36M     =       66957305        0       AAATGGCAAAAAGACAAAAATAAATAAATAAATAAA    ?????BBBDDDBDDDDFFFFFFIIIIIIIIIHIIII    RG:Z:0  NM:i:0  XT:A:R  md:Z:36
HWI-M00185:3:000000000-A1BH4:1:1101:5903:5381   145     chrX    66957305        254     36M     chr18   37526012        0       ACATGGTAACCTGTCTCATAGCAGGACTCTGGAATG    ?????BBBDDDDDDDDFFFFFFCHIHHIIIIIFHII    RG:Z:0  NM:i:1  XT:A:U  md:Z:1T34

Am I missing something or something is wrong here? How can read2's mate be on chr18 while it could be read1?

Reads from the two fastq files:

@HWI-M00185:3:000000000-A1BH4:1:1101:5903:5381 1:N:0:GTCCGCTTTATTTATTTATTTATTTTTGTCTTTTTGCCATTT
@HWI-M00185:3:000000000-A1BH4:1:1101:5903:5381 2:N:0:GTCCGCCATTCCAGAGTCCTGCTATGAGACAGGTTACCATGT

I used the GEM mapper
What do you think, bug or feature?
Thanks

[kmcarr]

In that SAM file, read1's mate is read2 but read2's mate is not read1. Is that possible? Both are on chrX, convergent orientation (-> <-), perfectly mapped (36M).

HWI-M00185:3:000000000-A1BH4:1:1101:5903:5381   353     chrX    66951034        254     36M     =       66957305        0       AAATGGCAAAAAGACAAAAATAAATAAATAAATAAA    ?????BBBDDDBDDDDFFFFFFIIIIIIIIIHIIII    RG:Z:0  NM:i:0  XT:A:R  md:Z:36
HWI-M00185:3:000000000-A1BH4:1:1101:5903:5381   145     chrX    66957305        254     36M     chr18   37526012        0       ACATGGTAACCTGTCTCATAGCAGGACTCTGGAATG    ?????BBBDDDDDDDDFFFFFFCHIHHIIIIIFHII    RG:Z:0  NM:i:1  XT:A:U  md:Z:1T34

Am I missing something or something is wrong here? How can read2's mate be on chr18 while it could be read1?

Reads from the two fastq files:

@HWI-M00185:3:000000000-A1BH4:1:1101:5903:5381 1:N:0:GTCCGCTTTATTTATTTATTTATTTTTGTCTTTTTGCCATTT
@HWI-M00185:3:000000000-A1BH4:1:1101:5903:5381 2:N:0:GTCCGCCATTCCAGAGTCCTGCTATGAGACAGGTTACCATGT

I used the GEM mapper
What do you think, bug or feature?
Thanks

Read 1 has multiple alignments. The FLAG value for the read 1 alignment, 353, indicates that this reported alignment is not "primary" meaning there is at least one other alignment for read 1. The primary alignment for read 1 is at chr18:37526012.

How primary/secondary alignments are determined and reported is a function of the mapping program used. I am not familiar with GEM mapper so can't provide any insight there.

[syfo]

Right, I forgot to precise that GEM is exhaustive and reports *all* possible alignments within a maximum number of mismatches. There are indeed many other alignments for read1.
Still, I find it surprising that the primary alignment is transchromosomal (chrX-chr18) when a perfect match is found for a mate on the same chromosome. I should probably ask the authors.
Thanks for your answer kmcarr.