paired-end assembly reverse transcribed RNA-Seqs

[mrjuggle]

Hi all,
i have a probably pretty easy question. Iam doing RNA-Seq right now using the MiSeq (2x250), after doing quality filtration iam struggling with assembling my paired-end reads. Not the assembly of already paired reads but really the assembly of the pairs of reads.

Assuming that we have two fragments of 250 bp and assuming that we set the minimal overlap to 20 bp, i would guess assembled reads are at least 460 bp long. What i see is completely different, i get a length distribution ranging between 220 (meanign shorter than single reads) up to ~ 500 bp. Why is this the case?!

Iam sure the answer is very easy but right now, i dont see the point.

Best
mrjuggle

[GenoMax]

How are you stitching your reads together? With software e.g. FLASH? Did your reads have adapters/in-line barcodes?

[mastal]

I'm not sure about RNA-Seq, but in general, the fragment length distribution depends on the library prep method, and size selection steps during the library prep.

[mrjuggle]

@GenoMax: software, PANDASeq / FLASH / COPE, barcodes etc. are already trimmed away

@mastal: size selection was done in a way that most fragmens were around ~ 400 bp (as recommended for MiSeq 2x250 bp)

[mastal]

the fragments are never going to be 100% the same length, you will always get some fragments shorter and some longer than the average.

[mrjuggle]

That's clear, the size range based on a HighSens Experion Chip was between 300-500 bp for all samples. Iam just wondering that's all. You would say the most likely reason is the size distribution of the library, is it?

[GenoMax]

That is correct. If the size selection had been done well then the spread of the insert sizes would have been tighter. It looks like you have examples all over the place. As long as the reads are overlapping reliably you are ready for the next step in analysis.