I had a Metagenomic 16s rRNA run on MiSeq v3 kit 2 x300 cycles. with good PF >89% but

[thermophile]

Yes, we tried it, although not extensively. It worked, just no better than without denaturation.

Remember there is nothing to stop the denatured molecules from re-annealing--especially at the adapter ends unless you include a denaturation.

--
Phillip

good to know, thanks

[pmiguel]

have you ever tried ampure cleaning after the denaturing step? That should work, right? Carboxyl groups would attract either single or double stranded. But maybe [PEG] would need to be changed?

Yes, we tried it, although not extensively. It worked, just no better than without denaturation.

Remember there is nothing to stop the denatured molecules from re-annealing--especially at the adapter ends unless you include a denaturation.

--
Phillip

[thermophile]

have you ever tried ampure cleaning after the denaturing step? That should work, right? Carboxyl groups would attract either single or double stranded. But maybe [PEG] would need to be changed?

[pmiguel]

interesting. I've had a few library pools with a drop off after 50ish bases but couldn't see primer diamers on the tape station (I don't routinely run my amplicon libraries on the tape station, just the problems). I'll keep this in mind for the future.

Just keep in mind you would likely need to run the samples after heat denaturation/snap cooling on a denature chip -- eg, an RNA chip.

--
Phillip

[thermophile]

interesting. I've had a few library pools with a drop off after 50ish bases but couldn't see primer diamers on the tape station (I don't routinely run my amplicon libraries on the tape station, just the problems). I'll keep this in mind for the future.

[pmiguel]

Phillip-what conditions have you seen primers annealed to full length amplicons? do you routinely heat denature then snap cool before cleaning amplicons?

Mostly from amplicon libraries submitted by customer labs. I'm speculating that they are primers and primer-dimers. But who knows? We have seen them in some Nextera XT libraries. And to less extents in other standard library types.

Heat denaturing prior to Ampure doesn't seem to impact the purification. Ampure for unknown reasons seems to be effective for removing small amounts of primers/primer-dimers annealed to full length library molecules, but not in cases where it looks like >50% of the molar contribution is from low molecular weight products.

Would be great to have a single-strand clean-up method that didn't require denaturing PAGE.

--
Phillip

[thermophile]

Phillip-what conditions have you seen primers annealed to full length amplicons? do you routinely heat denature then snap cool before cleaning amplicons?

OP- I look at %base to see what is happening in my amplicon runs. Does it look like the %base from your previous runs (all of which will be way out of normal Illumina spec)

[pmiguel]

Thanks a lot Phillip for your valuable inputs.

I have gone through your suggestions and looked into the data in greater details wrt possiblity of primer-dimer or adapter dimers. I have checked the fastq files with fastqc software to look for over-represented sequences but could not find the adapter-dimers or primer dimer sequences in that. SO i believe I can rule out the contamination of these dimers. On the other hand I have also checked out for over-represented samples if any of them have BA traces, I found none of them. I am planning to get BA profile for a few over-represented libraries done to get 100% sure that if there is any contamination in them (is it necessary to do BA on RNA-chip?)

I am attaching fastqc screenshot for your reference (same sequences are over-represented in all fastq files)

Those are just 16S sequences. Since FastQC only appears to be looking at the first 50 bases, I don't think it is a good assessment of the presence of primer dimers in your libraries. You could follow Genomax's advice. But it would be good to see the FastQC of the reverse read. If it is the reverse complement of that sequence then you either have a very short v-loop in the species you are sequencing, or you do have primer-dimers.

Do you have a service contract? If so, just ask Illumina for reagents to do a phiX run. This would show if your MiSeq had deteriorated severely from its normal specifications.

--
Phillip

[GenoMax]

You need to use a scan/trim program (I recommend bbduk.sh from BBMap suite) to scan and trim your data. You can't depend on FastQC to identify dimer contamination. Here is bbduk guide to get you started.

[AmitChaurasia]

The length of the reads in your .fastq file likely does not include low-quality base clipping. The MiSeq will happily call bases on background noise. So the length of the reads is not diagnostic for this purpose.

Your random samples that were run on the bioanalyzer -- did they include any of the ~dozen samples that contributed >2% of reads? Those would be the likely culprits for providing large amounts of primer dimers.

Even if these did not show primer dimers your samples may have included large amounts of primer-dimers annealed to full length library molecules. To detect these you would need to use some sort of denaturing electrophoretic size analysis. This could be denaturing PAGE, or if you prefer the Agilient Bioanalyzer, try:

denature your samples for 3 minutes at 95oC and then "snap cool" them in an ice-water bath. Then run them on a denaturing agilent chip -- we use the RNA pico chip for this purpose. Note that during the heat denaturation step you need some method to prevent your sample from just evaporating -- use of a heated lid thermal cycler is good for this purpose.

--
Phillip

Thanks a lot Phillip for your valuable inputs.

I have gone through your suggestions and looked into the data in greater details wrt possiblity of primer-dimer or adapter dimers. I have checked the fastq files with fastqc software to look for over-represented sequences but could not find the adapter-dimers or primer dimer sequences in that. SO i believe I can rule out the contamination of these dimers. On the other hand I have also checked out for over-represented samples if any of them have BA traces, I found none of them. I am planning to get BA profile for a few over-represented libraries done to get 100% sure that if there is any contamination in them (is it necessary to do BA on RNA-chip?)

I am attaching fastqc screenshot for your reference (same sequences are over-represented in all fastq files)

[pmiguel]

Hi Kinman,

Thanks for your reply, I have taken random samples to check with Bioanalyzer and no short peaks were found. Also fastq files were checked to have 301 base length only.

So this can be ruled out.

Best,

The length of the reads in your .fastq file likely does not include low-quality base clipping. The MiSeq will happily call bases on background noise. So the length of the reads is not diagnostic for this purpose.

Your random samples that were run on the bioanalyzer -- did they include any of the ~dozen samples that contributed >2% of reads? Those would be the likely culprits for providing large amounts of primer dimers.

Even if these did not show primer dimers your samples may have included large amounts of primer-dimers annealed to full length library molecules. To detect these you would need to use some sort of denaturing electrophoretic size analysis. This could be denaturing PAGE, or if you prefer the Agilient Bioanalyzer, try:

denature your samples for 3 minutes at 95oC and then "snap cool" them in an ice-water bath. Then run them on a denaturing agilent chip -- we use the RNA pico chip for this purpose. Note that during the heat denaturation step you need some method to prevent your sample from just evaporating -- use of a heated lid thermal cycler is good for this purpose.

--
Phillip

[AmitChaurasia]

Hi Kinman,

Thanks for your reply, I have taken random samples to check with Bioanalyzer and no short peaks were found. Also fastq files were checked to have 301 base length only.

So this can be ruled out.

Best,

[jdk787]

It looks like your Q scores have a big drop at ~80bp which indicates that you have a lot of smaller than expected library fragments. Since you are sequencing v3-v4 libraries I would guess this is due to amplified primer dimer, most likely in the libraries that got the higher read representation which you should see when looking at the data.

This wouldn't hurt the pass filter rate since that is calculated at read 25.
Did you visualize these libraries before sequencing?

[AmitChaurasia]

Hi All,

I had a metagenomic run on MiSeq using v3 kits of 16s rRNA v3-v4 region. I have loaded 8pM pooled library with 10% PhiX spike-in. MiSeq run had 89% PF with 16 M raw reads but with very low q30 37.5%. I am surprised how come q30 score that too much low even having good PF of 89%.
I also see a large variation in % reads representation for a few samples (21-samples) out of 257 samples. Could this result in low Q30 of MiSeq run, but how ?
I am attaching a few SAV screenshots for reference.
Thanks