duplicate level increase from MiSeq to HiSeq

[yub]

Hi, I did bisulfite seq on the same library, first using MiSeq to check quality, and then using Hiseq to get more reads.

But the duplication level was 50% from MiSeq, and 95% from HiSeq, even though the exact same library was used.

Was this increase in duplicate caused by natural duplication, or too much sequencing coverage? Are there any other explanations? Thank you very much!

[Brian Bushnell]

The most obvious answer would be that HiSeq produces more sequence, therefore the duplication rate must increase for the same library. How many reads were produced in each run?

[yub]

Thank you for answering my question. MiSeq generated ~2 mil reads, HiSeq ~100 mil. So does this mean that much less reads are needed, since the majority of them are going to be duplicates and thus removed?

Thank you!

[Brian Bushnell]

The more reads you generate, the more duplicates; but, you may also recover some low-abundance reads. Still, that library has very low complexity. I don't know anything about bisulfite's effects on library complexity, but the complexity generally depends on the genome size, amount of starting material, amount of amplification, and whether reads are paired or single-ended.

Depending on your experiment, your results don't necessarily mean you didn't sequence enough - they might mean that your library was just garbage in the first place, possibly due to overamplification or too little starting DNA. In which case, you need to revise your protocol and start over. So, what was your experiment, what was the genome, how much amplification was there, were the reads SE or PE, and how exactly did you quantify duplication?