Adapter-dimer or low library diversity?

[biancaneve0666]

Hi all,

I ran a multiplex of 50 WGS libraries on Illumina NovaSeq system and the data output is low, approximately 40% undetermined reads.

The input amount into the library preparation is <1ng each and the PCR cycles are approx. 18-24 cycles. There are a total of 10 samples x 5 replicates each.

This is the %base plot of the run. There is low base diversity in the first 25 cycles but not the remaining cycles.




Is this because of adapter-dimers? I have clean-up the adapter-dimer and it doesn't seem like it is due to adapter-dimers even though it has similar profile to what is indicated in Illumina website.

https://knowledge.illumina.com/library-preparation/general/library-preparation-general-troubleshooting-list/000001911

This is the library trace of my sample.


I have also checked the color balance of my index, they are well-balanced.

The sequencing provider re-sequenced my sample by pooling it with other researchers' samples, the data output has improved drastically.

Is this an indication of low library diversity? But the %base plot does not look like the low diversity base plot shown in Illumina's website.

https://knowledge.illumina.com/instrumentation/general/instrumentation-general-reference_material-list/000001543

[Tom Barker]

Hi biancaneve0666, I realise this is likely too late now but in case it is still relevant this looks like adapter dimer to me. The low diversity region in your % base plot is the right size for adapter dimer and then the %G stays elevated for the rest of the read which happens when you sequence past the end of a molecule and no fluorescent bases are incorporated as on a NovaSeq a G is called when there is no fluorescence signal. Also the diverse section in the middle corresponds to the index sequence. On a NovaSeq the % base plot for daapter dimer looks different in read 2 to read 1 and I've asked Illumina why this is the case when on older chemistries you would see a similar pattern to read 1 where you can literally read the sequence of the adapter from the plot but they weren't able to tell me why.

Although you can't see adapter dimer in your bioanalyser trace this is likely because you appear to have overamplification. This is when a PCR runs out of primers, the dsDNA denatures, and no primers bind, but as the molecules all have the same sequence at the ends (as they have adapters on them) they form bubble structures which themselves can also form daisy chains which appear as larger than they actually are when run through a gel. Adapter dimer that was present during the PCR can bind to other library molecules and appear larger than adapter dimer when run on the bioanalyser. One way to test this is to take an aliquot of your pool and perform a few extra cycles of PCR on it. This shouldn't be enough for the PCR to run out of primers so everything will be double stranded and appear as it's true size when run on the bioanalyser. Assuming the larger material on your trace above is overamplification then this shouldn't disappear and I suspect you'll also see an adapter dimer peak appear. It's worth noting that adapter dimer clusters very efficiently due to its small size so although you have for example 10% adapter dimer in your data, it might only represent a few percent of your pool by molarity.