16S Library Plant Tissue Troubles

[seanw]

Hi all I thought I would seek advice here as I have tried a number of tweaks and not getting the results I would expect. Here is the background

DNA was extracted from plant stems to allow for a survey of bacterial endophytes using 16S amplicons in the v4 region

Samples were extracted using a Qiagen DNeasy Plant extraction kit, samples were quantified and run on a gel to check for quality. Yes it will be mostly plant DNA but still, this was done.
Library preparation was done following the Schloss Wet Lab SOP as I have done numerous times before for soil, roots, and yes some plant tissue (mostly combo runs when plant tissue as involved).
Samples were pools and placed on MiSeq and the following stats were obtained on my latest v2 500 cycle (2 x 250bp) run.
@ cycle 244 I got
Density 722K/mm2
Passing Filter 72.2%
PhiX spiked at 9.2%
Qscore >=30 32.9% and this only went down to about 25% by the end of the run

This was not the first time to run this library as I did a nano flow cell run with it and had spike in PhiX and a much higher concentration to increase diversity and yet still ended up with a low pass filter and poor qscore.

I have, in the past run a library with these stem samples (a subset of 4 samples and two different extraction types Qiagen and MoBio on each of the 4 samples, to determine the best extraction method/bacterial yield) on a nano flow cell (2 x 250) and got
These stats:

Density 520k/mm2
Passing filter 88.5%
Qscore >= 93.6%
PhiX was at 13.6
Yes there was a fair bit of chloroplast/host DNA contamination but was getting about 12-20 % out after processing through Mothur.

I have also run other soil samples using the reagents and all are fine. I have also a large number of various library preps under my belt but am at a bit of loss on this one…

Any ideas?

[ATϟGC]

Hi Sean,

Aside from PhiX, are you taking any measures to increase diversity in your amplicon libraries such as staggering/offsetting and/or adding other loci?


At a glance, your cluster densities do not appear to be too high but if you have access to SAV (Sequence Analysis Viewer) for you run it might help if you posted the figures of %base/per cycle, density charts of the flow cell, and the density box plots. If not, the "per tile sequence quality" graph from FastQC might help show if there is differential quality over the length of the flow cell. Your amplicon run might not have been overclustered per se, but it might have been overclustered for a low-diversity run and 10% or more PhiX might not have been enough.

Here is a document on Miseq Overclustering in case you find it helpful:

http://www.well.ox.ac.uk/ogc/wp-content/uploads/2017/09/PreventingOverclusteringontheMiSeq.pdf

My guess is that your nano Mobio/Qiagen run was of whole-genome libraries rather than amplicons and may have worked better due to their high diversity.

I have seen people get lower PF% (~70%) with single amplicon sequencing despite moderate PhiX spike-in (~15-20%) if they do not stagger/offset their read 1 locus-specific primers.

[seanw]

Follow up info.

Hi,

here is some clarification on things.

There is no stagger within the primer pairs to increase the diversity and I have not added anything else, but this had not been an issue before (I guess it may be now). I also had a nano run of this library with a greater concentration(30-40%) of PhiX but that still had the issue.

The MoBio/Qaigen subsample library run was a 16S amplicon using the same primer sets that are giving me problem. It was a trial run I used to make sure this was doable with these types of samples.


Here are some screen shots and yes it certianly looks like a diversity issues...

[ATϟGC]

Hi again Sean,

Sorry for making that guess/assumption. I forgot that eukaryotes have 16S RNA as well.

So it does appear that you have very low base diversity, you have rapidly declining q scores that appear to coincide with increasing %G at base 120 of read 1, and most of your clusters passing filter come from lower densities (an indicator of overclustering). I am still leaning toward your troubles being from low-diversity and moderate overclustering but it could also be from bad reagents or something else.

This video goes over a lot of the same things as the document I added earlier:

https://www.youtube.com/watch?v=scU6vRLhnxE

I have lost one amplicon run to an expired kit which gave us a bunch of junk G nucleotides and rapidly declining q-scores. A re-run with the same library with new MiSeq reagents gave us excellent data. That being said, I have seen some colleagues who do not stagger/offset and/or mix in other loci get worse %PF and Qscores than I do with the same sequencing provider.

I have included 3 different offset/staggered amplicons in the 5 MiSeq amplicon runs I have done and they have all worked well with ~10-15% PhiX(aside from a 6th run that had the bad reagents). Perhaps someone with more experience interpreting SAV results will be able to tell you which problem your results are indicative of.

Some screen shots from FastQC would likely help with diagnosis as well.