NextSeq 500, High-Output Paired end reads - upper/lower limits for insert sizes?

[husseh01]

What is the upper/lower limits of the final library template that can achieve an optimal clustering efficiency & success rate? Illumina states that an "insert size of 550 bp is supported" but they don't give any range limits before sub-optimal clustering efficiency occurs. Online NGS forums have stated anywhere between 200-800 bases, but this is not supported by any relevant data -can anyone confirm this? We currently use the NextSeq 500, High-Output Paired end reads.

[Genetic Librarian]

we have successfully clustered 1000bp libraries (final size including adapters). You will end up with less cluster passing filter, but we were still within specs. On the lower end, I have no experience.

[nano85]

Library with 1000bp fragment size on NovaSeq

we have successfully clustered 1000bp libraries (final size including adapters). You will end up with less cluster passing filter, but we were still within specs. On the lower end, I have no experience.

Hi Genetic Librarian,

I have a TruSeq Nano library with 1000bp fragments (final size with adapters - see BioAnalyzer trace attached) that I want to run on the NovaSeq. I realize that the NextSeq and NovaSeq platforms behave differently, but were there any modifications that you made to your sequencing run to accommodate 1000bp library fragments on your NextSeq platform?

Thanks for any guidance!

Nathan

[pmiguel]

Illumina instruments preferentially cluster shorter amplicons. The Agilent chip you include shows visible material below 600 bp. Since this lower molecular weight material is present, it will displace longer amplicons and you will see zero, or close to zero, 1000 base insert results.

--
Phillip

[nano85]

Previous large-fragment-sized library and insert sizes

Hi Phillip,

That jibes with what I've heard about cluster generation bias on Illumina platforms as well. However, just to verify this, I checked the insert size distribution of reads from a library with a similar fragment size distribution (mostly 1000 bp) that I had sequenced on a 300-bp-paired-end MiSeq run last year. My insert sizes (calculated by bwa and extracted from the resultant sam file: see attached pdf) mostly clustered around ~550 bp. If I correctly understand how insert sizes are calculated, then my fragment sizes could be calculated by adding insert size + 2 * read length, or 550 + 2 * 300, which is 1150 bp. Is that correct? If that is correct, then I was able to get somewhat plentiful clustering and sequencing of ~1000 bp fragments (on the MiSeq system, at least).

[GenoMax]

@nano85: I suggest that you actually calculate the insert sizes by using one of the methods noted by @Brian in post #2 here.

[nano85]

@nano85: I suggest that you actually calculate the insert sizes by using one of the methods noted by @Brian in post #2 here.

@nano85: I suggest that you actually calculate the insert sizes by using one of the methods noted by @Brian in post #2 here.

Hello, Phillip. Thank you for the advice. I did as you suggested (following method one from post #2 by Brian Bushnell: map reads to an assembly) and got the following insert size stats, which confirm the results from bwa:

insert size avg: 585.29
insert 25th %: 478.00
insert median: 567.00
insert 75th %: 664.00
insert std dev: 154.24
insert mode: 565

However, I was calculating the fragment size incorrectly: fragment size = insert size + 2 * adapter size (read sizes are included within the insert size). Adapter size with an index is 64 bp, so my average fragment size is 585 + 2 * 64 = 713 bp. This supports what you said earlier: that smaller clusters are sequenced more efficiently. The library does have a lot of sequenced fragments on the large side, though.

[Genetic Librarian]

You also have to keep in mind that the flowcells and clustering methods are not the same in the NovaSeq as in the MiSeq / NextSeq. A patterned flowcell is much worse for sequencing long libraries, since clusters bridge into neighbouring wells.
Also, the ExAmp chemistry has a higher bias against long fragments as the traditional bridge amplification.