[pmiguel]

Two weeks from now, I'll run some bacterial total RNA libraries which shouldn't have too much of an issue with diversity. Maybe I'll try one the "Illumina way" and one the "pmiguel way" and see what the results will bring. Have to think about this.

Perhaps the "Purdue Way" rather than "pmiguel way" would make me sound more modest?

--
Phillip

[dfhdfh]

Haha, ok

I'll report as soon as I know more.

[dfhdfh]

Ok, these are the results:

First run (1x150 v3, high diversity sample), I spiked in 5 % PhiX (Illumina way), which resulted in 3.85 %. As I was quite happy with this result, I performed the second run (also 1x150 v3, high diversity sample) with 1 % spike-in (also Illumina way), which resulted in 0.8 %.

Today, I started an amplicon run (2x300 v3) with a spike-in of 10 % and used the same PhiX lot as for the samples from last week (Illumina way). However, this time I only achieved 5.02 % of PhiX (it's still running and only 50 cycles in but I don't think it'll change).
Well, that's rather disappointing and I have no clue what the difference is. Somehow the library seems to have an influence? I'm not sure.

[microgirl123]

The percentage of phiX you end up with is also dependent on how accurate the quantification of your actual library is. If your library is underquantified, you will end up with a lower percentage of phiX. If it's overquantified, you'll end up with a higher percentage. That's why I usually aim for a higher percentage than 5%. I'd hate to lose a low diversity run over the phiX concentration.

You also need to keep in mind that you are pipetting really small volumes, both in loading your library and in quantifying it. Even with perfect pipetting skills, it's impossible to be totally accurate. One thing that I find helpful is, if I end up with a very high concentration library (based on the KAPA quant), to dilute it down to less than 10 nM and quantify again before loading the MiSeq.

[dfhdfh]

That's a really good point. However, the libraries were all quantified in the same way and I can predict pretty well what cluster densities I'll get. So I'm actually quite confident with my measurements.

[pmiguel]

That's a really good point. However, the libraries were all quantified in the same way and I can predict pretty well what cluster densities I'll get. So I'm actually quite confident with my measurements.

Yeah, this is what we were seeing. Crazy variation in the number of phiX clusters.

--
Phillip

[Jessica_L]

Might it be an issue with either the NaOH or the PhiX library's buffer? Maybe the concentration of the buffer (or one of the components) is off even if the amount of PhiX DNA in the tube is correct? That sort of makes sense in my mind since mixing the PhiX and main library (the Purdue way) might dilute whatever inhibitors are present down to insignificant levels.

When I've seen cluster numbers deviate from expectation, they were usually related to the pH of the sodium hydroxide. I've taken to checking with pH strips before starting my denaturation step. I've had to toss more than one tube of NaOH on more than one occasion.

I'm not sure that either of these explanations adequately explain why you're seeing differences at high percentage vs low percentage spike-ins, though. Denatured PhiX should be denatured PhiX and it would either work or it wouldn't.

Might there be some sequence in your bacterial libraries that could be interacting with the PhiX sequence at the flow cell hybridization step?

[pmiguel]

I like the idea of phiX library's buffer being the culprit ...
Seems like we never have trouble with the HiSeq in this regard, though.
Also doesn't explain dfhdfh's results.
--
Phillip

[Jessica_L]

That was exactly my thought-- it doesn't really explain the entire issue since a lot of the variability seems to be instrument dependent. Maybe it's somehow related to onboard clustering? In which case, a HiSeq in rapid mode should have the same issue?

[GenoMax]

Clearly people take phiX seriously

Does it matter if one recovers less (based on other posts), as long as it positively influences a run in reaching completion?

[dfhdfh]

In my mind it does, because, on the one hand, I want as little PhiX as possible in order to maximize my actual data output. On the other hand, I want a high enough percentage in order to enable sequencing of low diversity samples. In theory, this balance is simple, in practice it is a bit of a pain.

[sweetph3]

What is the fragment size of the PhiX library? I haven't ever run it on a Bioanalyzer, so I don't know if it's relatively small or large. Since smaller fragments tend to cluster with better efficiency than larger fragments, could the variations seen here just be due to certain libraries clustering more (or less) efficiently than the PhiX controls because they are smaller (or larger) than the PhiX?

[martagh1]

I have had these issues with PhiX not aligning as well as it should when I rum miRNA libraries in either MiSeq or NextSeq, and Illumina has told me that it is very dependent on the size of your fragments. Sweetph3 you are right, smaller fragments cluster better than larger ones, so if your library has an average size of 150nt (like the miRNA ones) they will cluster way better than PhiX which is larger (if I remember correctly Illumina told me it is around 465bp). So you have to make sure, in those cases, that you load more PhiX than normal to make sure it aligns closer to the expected spike-in, and as such you get good enough diversity.

[Jessica_L]

From Illumina's technote:

"The mean insert size of the PhiX v3 library is approximately 375 bp,
corresponding to approximately a 500 bp library size if visualized on a
Bioanalyzer."

[dfhdfh]

So, in my case, PhiX should cluster better than my library, whose size (~630 bp total) I exactly know because it's an amplicon library. However, that's not the case.