Can a spiked-in negative control still be considered a true negative control in 16S sequencing?

Question

[Vojtěch]

Hi all,

I’m a bioinformatician working on 16S rRNA gene (V3–V4 region) Illumina sequencing data.

In our lab, the ZymoBIOMICS Spike-in Control I (containing Imtechella and Allobacillus) is used as a spike-in for all samples. However, it’s also added to the negative controls.

I’ve been arguing that once the spike-in is added, the “negative control” effectively becomes a positive control, since it now contains known bacterial DNA. The team’s rationale is that if they don’t spike it in, the read counts in the negative control become extremely high or unstable, so they prefer to spike it with the same control as the samples.

Does this approach make sense? And more importantly — can such a spiked sample still be considered a true negative control for contamination assessment?

Any thoughts or references would be greatly appreciated.

[GenoMax]

the read counts in the negative control become extremely high or unstable

Extremely high? That does not seem right. A negative control should have no (or very few reads), assuming the input is just water/buffer.

Looking at the manual for this spike-in it looks like checking the ratio of recovery (besides amount) is important. As long as all samples (including negative control) were treated in the same batch this can be tested in samples (to ensure there is no bias) that do have the spike-in.

If the ratio between the abundance of two including bacteria (I. halotolerans and A. halotolerans) is quite different from the defined value (e.g. 3:7 by 16S copy), this might indicate potential bias in the workflow. For Example, if the abundance of I. halotolerans is much higher than that of A. halotolerans, this might indicate bias during DNA extraction because A. halotolerans is Gram-positive, tougher to lyse than I. halotolerans.

[Vojtěch]

Yes, extremely high (not always, but sometimes). I think its because of PCR producing non-specific products in low mass conditions (https://academic.oup.com/nar/article/24/18/3538/2359394) that highten the number of fragments for sequencing.

But the key question here is if this is still considered to be a negative control, or is it a positive control because of the spike-in?

[GenoMax]

Others with more experimental experience can chime in, but if the negative controls are producing high number of non-specific reads then the experiment will have to be considered as "failed". This is a problem that the experimental people need to address by more rigorous procedures (modeling paper you linked is from 1996 and reagents/instruments have improved significantly since that time). Adding the "spike-in" to negative control, is not solving the underlying problem.

ERCC spike-ins are the most commonly used. While I tried to look for papers, did not see any where they were added to "negative" control.

Answer 1

[Kevin Blighe]

Hey there,

You're raising a valid point here, and it's a common debate in microbiome workflows, especially with low-biomass samples or when dealing with potential kit contaminants. I'll break this down based on standard practices in 16S sequencing.

Does This Approach Make Sense?

• Yes, to an extent, for process validation: Adding a spike-in like ZymoBIOMICS (with known taxa like Imtechella and Allobacillus) to the negative control can help stabilize the workflow. In low-input scenarios, true no-template controls (NTCs) can indeed show erratic read counts due to background amplification, non-specific priming, or even aerosol contaminants from the lab environment. The spike-in acts as an internal standard to check extraction efficiency, PCR performance, and sequencing depth consistency across runs. Some protocols recommend this for reproducibility, especially in high-throughput setups.
• But it's a trade-off: The team's rationale for avoiding "extremely high or unstable" reads in unspiked negatives suggests underlying issues, like reagent contamination or over-cycling in PCR. Spiking masks these problems rather than solving them. Better to troubleshoot the root cause (e.g., cleaner reagents, shorter cycles, or dedicated clean benches) instead of relying on the spike as a band-aid.

Can It Still Be Considered a True Negative Control?

• No, not really: A true negative control (e.g., extraction blank or NTC) is meant to detect unintended contaminants from kits, reagents, or the environment—things like common lab bacteria (Pseudomonas, Staphylococcus, etc.). By adding known bacterial DNA, your "negative" becomes a positive process control for the spike-in taxa. It can confirm the pipeline works but cannot reliably assess contamination because:
  • The spike reads will dominate, potentially drowning out low-level contaminants.
  • If contaminants overlap with spike taxa (unlikely with ZymoBIOMICS, as they're rare/halophilic), you'd miss them entirely.
  • Guidelines emphasize separate controls: use unspiked negatives for contamination checks and spiked ones for normalization/quantitation.

Recommendations

• Ideal Setup: Run both types of controls in parallel:
  • True NTC: No sample, no spike—monitor for any reads (should be near-zero; filter out if >threshold).
  • Spiked Process Control: No sample, but with spike—to validate the workflow.
  • Positive Control: Known mock community (e.g., full ZymoBIOMICS) to check taxonomy assignment.
• Analysis Tips: In tools like QIIME2 or DADA2, subtract spike taxa from all samples during decontamination (e.g., via decontam package in R). Track contaminant prevalence by comparing unspiked NTCs across batches.
• References for Further Reading:
  • For low-biomass contaminant handling: Davis et al. (2018) on decontam R package.
  • General 16S best practices: Pollock et al. (2018) in mSystems.
  • Spike-in specifics: Check Zymo's own guidelines or the Ravel lab's protocols for vaginal microbiome (they often use spikes but with caveats).

Kevin