QV distribution

[eoh001]

hello
I have an interesting distribution of QV from the whole transcriptome reads from a gastric cancer cell line.
The librariy was prepared by the multiplex paired-end sequencing kit(16 barcodes). 3F(50bp), 5F(35bp) and barcode(5bp) sequences were obtained with SOLiD 4 system.
I checked the quality values of the reads by base position, and I observed that the 3F had a weard cyclic fluctuation.
The QVs of all positions in the 3F reads were very poor except (2, 7, 12, 17, 22, 27...42,47.) positions.
The 5F and barcodes showed good QVs across all positions, although they had 5-base cyclic fluctuation across a read due to the primer reset.
I think something wrong happened with 3F sequencing. What could be the reasons on the poor QV in the 3F reads?. Any one can explain the observation? I'll really appreciate any ideas. THANKS.

[Brian B]

Solid reads using 5-bp long fluorescing thingies. They get attached one at a time until the entire end is full, then they are removed, finishing the cycle. Then offset is changed, and they are placed one-by-one again. So you might read like this:

1, 6, 11, 16 ... 41, 46
then remove them, and read
2, 7, 12, 17 ... 42, 47
etc.
This is done for 5 cycles so that every unique position is read once.

So, the quality values should be most consistent for every set of positions that is equivalent modulo 5, and should be highest for the first cycle and lowest for the last cycle. In practice I don't really observe that, but I have not looked for it rigorously.

[asr]

The QV plots of SOLiD data do have that saw tooth shape for the reasons outlined previously. Normally, you would see a decline of up to 5 points then reset to something slightly lower than the original score. I'm afraid to say that your F3 read looks quite poor.

However, your F5 reads looks good. Is this with the new chemistry as it is much better than anything I've see with the original PE kits?