High duplication

[kmin-ryu]

My samples showing high duplication rates in MultiQC reports.
I know that QuantSeq library is inherently low complexity, so I understand mild levels of duplication is fine.
But some samples are 'more duplicated' than others... would it mean that the library is PCR overamplified?
Those samples become outliers in PCA plot, too.

Do you think using UMI at the library prep would help?
(https://www.lexogen.com/store/umi-second-strand-synthesis-module-for-quantseq-fwd/)

Thank you!

[t-neumann]

Hi,

indeed QuantSeq is inherently of low complexity. We tried it with the UMI library prep also, where we saw duplication rates of around 60%, but the overall readout was very well correlated with the non-deduplicated reads, so generally speaking we didn't observe any major biases introduced with duplications so far.

[kmin-ryu]

@t-neumann
Thank you!

But do you think ~90% of duplication rates might be too high, maybe?
Does the alleyoop dedup command perform deduplication considering UMI?

[t-neumann]

No unfortunately alleyoop dedup does not consider UMIs, just same start/end positions. What's the percentage of reads after the slamdunk filter steps that are left? Regardless of being claimed to be duplicates or not

[kmin-ryu]

@t-neumann

Oh, so if I want to introduce UMI than I should use some other tool, too.

MultiQC report shows like this.

[t-neumann]

Sorry for missing this one - do you also have the stats sequenced / mapped / filtered / counted like from the top of the multiqc report?