Read length difference between the analysis type and true reads

Question

[ognjen011]

I have noticed a long time ago that the true read length may not exactly match the promised read length in the analysis name e.g. 2x100bp Illumina data frequently consists of 101bp reads. Today I looked into a dataset that systematically has 97 or 96 bp reads for the promised 2x100bp across many files. Why does this happen and should it be interpreted in any way?

Thanks in advance for the comments!

Answer 1

[mastal511]

The explanation for having one base more than expected (e.g. 101 instead of 100), is that for Illumina data, the phasing for each position in the read depends also on the data from the next base, so the last base in the read will always have a much worse base quality, because there is no data from a next base. So by having 101 instead of 100 bases you should get 100 bases with reliable base qualities.

For the cases where there are 96 or 97 bases, I'm guessing that either the run ran out of reagents, or else the base qualities at the 3' ends of the reads were poor and someone has already trimmed the last few bases.

[GenoMax]

To add to this: some sequencing facilities run n+1 cycles to get you n cycles of data with reliable quality. This may be more critical for the tag reads than the main.

[ognjen011]

Sorry, I tried to find it but failed - what is the difference between tag reads and main reads?

[GenoMax]

Main are reads contain sequence from your own samples. Tag/index reads are the short oligos used to "code" individual samples (during library prep) so multiple samples can be pooled together and run in one lane. Based on the sequence of these tag reads, it is possible to bin reads belonging to individual samples after the run is done. In Illumina technology, tag reads are sequenced independent of the main reads.