Lib prep - Splitting PCR to reduce PCR bias

[prepagam]

I want to reduce my high levels of duplicate reads (>50%) in my illumina HiSeq 100bp Paired end sequences.
The normal solutions aren't possible as I am using the max DNA available per sample (80-200ng), and only with 10-12 PCR cycles, my library is strong enough to sequence.

I wondered if I could reduce PCR bias by splitting the PCR into two tubes, and then pool after PCR? This idea is based on the assumption that the bias generated by PCR is random, and therefore different PCRs will over/under represent different parts of the genome.

A similar approach is used in some protocols for RADseq - but their library prep does differ.

[kcchan]

How much library are you getting with 10 cycles of PCR? 100ng of DNA should get you plenty of library for sequencing. You should double check your upstream steps to ensure everything is working efficiently, especially the adapter ligation.

If you're running the cBot instruments internally, there are also tricks to load your samples starting with less than 1nM of DNA library.

Another question is how you're determining the duplication rate? Some programs such as fastqc do not accurately determine the duplication rate.

[kmcarr]

What is your source DNA material?