strange SAM output

[zhidong]

Hi,

I use tophat to run RNAseq project, the sequence is from solexa pair end .
the below is part of one line in the result.

ILLUMINA-7A0261_0001:3:77:8496:8665#0 161 chr1 554335 3 64M = 1395011 0

from the SAM1 manual, I think the insert size is 840676(1395011-554335). but ISIZE(inferred Insert SIZE) is 0. how can explain it.




zhidong

[macrowave]

Can you try using other short read mapper such as bwa to see the insert size distribution? Or you can use the Bio:B::Sam perl modules to access the entire sam/bam, and infer the insert size distribution.

[zhidong]

thank you , macrowave.

but I run RNAseq project. bwa seems not fit for mapping RNA sequence to Genome

[macrowave]

Can you provide more specific information, such as the fragment size, read length, the expected insert sizes, reference type (genome or transcriptome)? Form my experience, paired-end mapping with bowtie to predicted transcriptome yielded expected insert size distribution. By the way, BWA is perfectly fine for mRNA-Seq mapping to the genome, it's just harder to estimate insert size because of the introns, and you'll get weird inferred size distribution as the variable intron length. The Bio:B::Sam perl module has functions to access all proper mapped paired reads from sam/bam , from no matter which mapper you use. So it's a good idea to get all pairs in a region and see the real distribution.

[macrowave]

Just realized that your problem might be a bug in TopHat. In the newest TopHat release notes, they say 'TLEN field in SAM format is correctly output', which means you may be using an older release that doesn't output the isize correctly.

[macrowave]

and the sam flag 161 (1+32+128) means the paired reads mapped one forward, one reverse, but for some reason, the aligner thinks the pair isn't right (not properly aligned?), so that might be also a reason it returns a zero isize.