Spiked-in control for microbiome

[[email protected]]

Hello! I am helping out with a microbiome project and my collaborators have had a problem that can best be explained by a sample naming error at the sequencing facility. They have agreed to resequence our samples but we have lost some faith! We would like to add an internal control to some samples to be able to distinguish them. I am thinking of adding some genomic bacterial DNA that would not be found on our environment to some, but not all, of the samples.

My question is technical. I wonder how much to add, and if adding a known amount of DNA to our environmental samples can affect the results in any way. I couldn't find any protocol for spiking environmental samples with bacteria for this purpose.

Any advice on what to use and how much to add would be much appreciated!

[Brian Bushnell]

I suggest using custom synthetic oligos for spiking that could not possibly compromise your results. ~1% should be good, though it depends on how many reads you are expecting per library.

Can you give a little more information about the degree of multiplexing, number of reads expected per library, and so forth? And given that this is 16S, do you need to spike in bacterial DNA to match the amplification primers, or are you sending them already-amplified samples, in which case you could spike in something else?

[[email protected]]

Thanks, Brian! We are providing DNA extractions to the sequencing facility, so I would want to add something that would be amplified by their 16S primers. But, I could create some synthetic oligo that includes the primer sites. The facility guarantees 10,000 reads per sample, with a target of 15,000. For the first round of sequencing we received anywhere from 8,000-25,000 reads. I am not sure about the degree of multiplexing, but I would guess high given how few (relatively) reads per sample we are aiming for.

[Brian Bushnell]

If you create some synthetic oligos, you could standardize on spiking them into everything with no worries about incurring analysis problems, since they'll be very easy to remove, and that will make it easier to quantify, which allows checking for cross-contamination in addition to sample mixups. It could even improve your sequence quality by increasing color diversity (though that's hardly the goal, or an expected outcome at such a low level of spiking). For 10k reads 1% should be sufficient, at least to find sample mixups, if not low-level cross-contamination. Normally when we spike tracer sequences into a plate, only half the wells get it and the other half don't, to save on the number of unique sequences needed. Though you don't really need a unique sequence for each well; statistically, you'll be able to detect most major problems with a handful of different sequences.

[thermophile]

Have you heard an adequate explanation for what happened that leads you to believe that it wouldn't happen again? Do you know how long it will take to get your new data back? I'd want those questions answered before sending more samples.

But for your question, rather than getting synthetic DNA made, I'd add some extra control samples (pure cultures or defined communities).