OverClustering - MiSeq v2 reagents - Nextera XT Library Prep

[clseq2017]

Hi everyone,

I am a new user of MiSeq. I am using the Nextera XT Library Prep for sequencing of gDNA along with the MiSeq Reagents v2.

I am having issues with overclustering, and the Q30 was pretty low (58.3%). However, the Cluster Passing Filter % was still around 79%. The images did not look terribly overclustered tough.

I have heard that reviewing parameters on Reads 1, 2, 3 and 4 can help understand what happened but I don't know how to interpret those parameters.
I noticed that for Reads 1 and 4, the Q30 is low : 63%, and 51.6% respectively. However, for Reads 2, and 3, it's pretty good: 90.4%, and 90.8% respectively. What does that mean when the error rates is high for Reads 1 and 4???

Thanks so much for your help!!!!

[finswimmer]

Hello,

Read 2 und 3 are the index read. As they have just a few cycles the Q30 is alway higher than in Read 1 und 4.

What was the Cluster density? Have you had better runs before? What kind of gDNA do you try to sequence (human?) ? Maybe you have low diversity in your target regions, which makes it hard for the MiSeq to distiginguish the clusters . Spike in phiX might help.

fin swimmer

[hopedlamini]

Hi everyone, I have sequenced bacterial DNA using Illumina MiSeq, and have both forward and reverse reads, after quality trimming, the forward read has high GC content, is there any software of trimming this? Thanks

[clseq2017]

Hello,

Read 2 und 3 are the index read. As they have just a few cycles the Q30 is alway higher than in Read 1 und 4.

What was the Cluster density? Have you had better runs before? What kind of gDNA do you try to sequence (human?) ? Maybe you have low diversity in your target regions, which makes it hard for the MiSeq to distiginguish the clusters . Spike in phiX might help.

fin swimmer

Hello,
The cluster density was very high 1600 K/mm2 (supposed to be 1000-1200 K/mm2 for V2 reagents).
Yes, I have had better runs but the cluster density has always been in the upper limit. I am sequencing bacterial gDNA from several species.
Maybe this is low diversity. How can I double check that in the run parameters?

Thanks!

[clseq2017]

Nucleotide Diversity

Hello,

You mentioned "low diversity" library but looking at some troubleshooting guidelines it looks like my library is unbalanced (see attached picture).
What is an unbalanced library? How can I fix it?

Thanks!

[GenoMax]

You are way over spec for the cluster density. I am surprised you were able to get any data at all. I would not trust the data you have if it is to be used for any type of de novo work. You should think about re-running this at half concentration.

[clseq2017]

You are way over spec for the cluster density. I am surprised you were able to get any data at all. I would not trust the data you have if it is to be used for any type of de novo work. You should think about re-running this at half concentration.

I agree with you! But Illumina looked at the run and they said the parameters are still acceptable. However, I want to repeat and fix this overclustering issue. But my questions are : how do I avoid overclustering???? and how do I lower the loading concentration? since I am using Nextera XT that does the bead-normalization.

Also, I checked the pooled library and it looks like there is nothing showing up! so I am very confused!!!
Our bioinformatician said "The 3’-end quality is quite poor for both the R1 and R2 and there are strong 3’-end sequence biases".
Why is the quality poor on 3'-end? Any explanation?

Sorry for all those questions, but it's my first MiSEq failed run!!!

Thanks!

[finswimmer]

Hello,

to your question about how to avoid overclustering: What was the concentration of your Library at the end of of the LibPrep? Have you measure it with qubit? What is the average fragment size? And with how many pM have you load the flow cell?

fin swimmer

[Genetic Librarian]

and how do I lower the loading concentration?

During the denature and dilute process you dilute your library to the final loading concentration with the provided buffer based on your final Qubit quantification of the library pool. So, instead of diluting it to e.g. 10 pM loading concentration, try 6 pM next time.

[GenoMax]

I agree with you! But Illumina looked at the run and they said the parameters are still acceptable.

It is interesting if they actually said that. Perhaps they did not want to spook you

Our bioinformatician said "The 3’-end quality is quite poor for both the R1 and R2 and there are strong 3’-end sequence biases".
Why is the quality poor on 3'-end? Any explanation?

That is to be expected. When a run is overclustered the clusters are very close to begin with. Over the run they become fatter and then the software has a tough time telling them apart. You could easily lose a run due to this. In your case though since this is mix of genomic DNA's the run was able to end successfully but the Q-scores suffered when the reliability of basecalls dropped towards the end of the run.