This is adapted from a question posted on reddit

In my work, I occasionally get asked about metagenomic sequencing, which (after further questioning) I discover means that the person wants to work out the taxonomic composition of a sample.

In the past, people have gone straight to 16S sequencing for this, and various manufacturers have created targeted 16S kits for this purpose. But does it still make sense to do that? Is 16S still cheaper and better than other options?


1 Answer 1


When comparing anything targeted to whole-genome sequencing for the purposes of assembly, the targeted approach will always be cheaper, because the total length that needs to be assembled is shorter. But that's like comparing a video to a photograph.

If you're just looking to determine the phylogenetic composition of a bacterial sample, then there's no reason why targeted 16S sequencing should be cheaper than whole-genome shotgun sequencing, and the whole genome approach may end up better due to the highly conserved nature of the 16S sequence, and the abundance of horizontal gene transfer.

Previously, 16S might have been better because the taxonomic databases were more complete, but now that there are so many microbial genomes available, the whole genome databases are comprehensive enough for most purposes.

@terdon linked to this 2008 paper, in which they compare random sequence reads to 16S for estimating the biodiversity of a metagenomic library, finding that 16S wasn't worth it and they could get perfectly good results with random reads.

In a situation where samples are abundant and no amplification is required, PCR can be excluded from whole-genome sequencing, making it the cheaper option at the sample preparation side of things.

Just to put numbers to things to demonstrate how cheap whole-genome sequencing can be, here are the big consumable costs for nanopore (prices in USD):

  • RBK114.24 rapid barcoding kit* - \$700 (24 samples, 12 runs)
  • Flongle flow cell - \$810 (12 flow cells)

So the marginal cost of one run is \$68 for the flow cell, and \$60 for the sequencing kit, processing 24 samples at a cost of about $11 per sample (rounding up).

[FWIW, ONT also provide a rapid PCR barcoding kit for the same cost (i.e. \$700, but polymerase is not included), which can be used together with PCR for low-yield samples.]

Flongle flow cells will usually generate >200 Mb per run (up to about 2 Gb in rare cases), so that's about 8 Mb of sequence per sample, which should be sufficient to do proportional counting down to about 1% of the sample (assuming >1000 reads per sample).

* There is a 96-barcode kit available for \$990, but I don't think that would have enough coverage to be useful when running samples on a Flongle flow cell.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.