Two sizes in sequence length

[jetjr]

I'm working with Illumina Miseq data. After BAM to fastq extraction I run fastqc to generate reports. With many of my samples I see two different sequence lengths as shown in the picture below:



Also want to add that if I see this in the sequence length distribution, it also means the sequence composition will be skewed:



Any idea what is causing this?
Single End Data
Adapters are removed off the sequencer

Thanks!

[jhalpin]

... you have single end Illumina data? Is that a thing?

We use Ion Torrent technology, but if I see that it's usually a size selection problem.. something went wrong with my bead ratios.

[Brian Bushnell]

This is odd. Illumina does not support read lengths greater than 300bp as far as I know (and even 300bp lengths are not well-supported), so I'm not sure what you are doing.

Can you explain the library-prep and sequencing methodology in more detail?

[GenoMax]

It is possible to run strange lengths e.g. 2 x 300 kit can be run as 600 single-end. Not sure why anyone would want to but one can.

@jetjr: Can you attach a complete fastqc analysis file?

[nucacidhunter]

In addition to questions in above posts, I wonder if these plots are from the same sample or file. Sequence length distribution plot shows a flat 0 count on Y axis for sequence length over ~275 while Per base sequence plot indicates that read length are up to 370 base.

[pmiguel]

Probably from merged reads from a 500 cycle run. (The plots are from fastqc, not SAV.) The shorter peak from reads that didn't merge.

BTW, can I add that fastqc's variable width bins make me feel uneasy?
--
Phillip

[GenoMax]

BTW, can I add that fastqc's variable width bins make me feel uneasy?
--
Phillip

Easily fixed by adding --nogroup to command.

[jetjr]

Thanks all for the replies! I stand corrected it is Ion Torrent data >.< We are looking into some of the things mentioned in this thread. I will add full QC reports soon...