[sarvidsson]

The k-mers in the ends of the reads could be short adapter remnants not identified by the adapter clipper. Typically they get soft-clipped by the aligner, but you should get rid of them if you use BBDuk in trimbyoverlap mode.

[rozitaa]

Mammals have teeny little exons spread out over 10's-100's of kilobases of the the genome. Mapping RNA (which has the introns spliced out) reads to the genome isn't a good way to determine insert size. And only getting 50% of the reads to map "concordantly" doesn't seem so bad. How is bowtie2 going to handle reads spanning a splice site?

If you want to determine your insert sizes, try aligning your reads to a long (spliced) transcript instead of genomic DNA. In my experience with the MiSeq and HiSeq, your sizes will look like all the very shortest library products were sequenced preferentially.

--
Phillip

Thanks Phillip! I think it should be like that since now that I am aligning only read one I can get approximately the whole aligned!!! Sure I will try to align it against a transcript reference.

[Brian Bushnell]

I think the problem lies in the bowtie2 command:

-1 R1.paired.fq -2 R1.unpaired.fq

Based on the filenames, it appears that you are telling it to treat read1's as pairs with each other, using the output of something like Trimmomatic that produced 4 output files. You should be doing something like this:

-1 R1.paired.fq -2 R2.paired.fq

Although, I'm not really sure why ANY of the pairs were concordant.

[dpryan]

I think the problem lies in the bowtie2 command

Good catch! I share your wonderment that any were concordant. I'm also surprised it didn't throw an error that the files were of different lengths.

[rozitaa]

I think the problem lies in the bowtie2 command:



Based on the filenames, it appears that you are telling it to treat read1's as pairs with each other, using the output of something like Trimmomatic that produced 4 output files. You should be doing something like this:



Although, I'm not really sure why ANY of the pairs were concordant.

Oh no this is not the problem sorry for confusion. This is my fault in naming the files incorrectly in previous step! The R1_unpaired.fq is supposed to be named R2_paired.fq!!!

[rozitaa]

Good catch! I share your wonderment that any were concordant. I'm also surprised it didn't throw an error that the files were of different lengths.

I might have caught up the problem when I align it with only single end reads (readI). These are RNA-seq data that I am trying to map against genome reference. with bowtie I cannot get the splice junctions! So for getting the inner distance by this way I should align my RNA-seq reads against a transcriptomic reference (this mentioned before also).
Anyway now I almost have my inner size and I am trying to align them by TopHat. I will update this thread by the result of my TopHat aligning.

Thanks all!

[rozitaa]

Thanks everyone! Now it is aligning perfectly fine with TopHat! to sum up, I got the reads inner size from aligning firstly against transcriptomics reference by bowtie and then I aligned against genomics reference by TopHat!