Demultiplex 16S hypervariable regions

[colinbrislawn (Colin J Brislawn)]

Good morning,

@apc

But, is there any Standard Operating Procedure to perform this task with QIIME?

Not yet. How to handle multiple hypervariable regions at once is still an open question.

I noticed you mentioned the Ion Torrent 16S kit and its secret primers. I know other members of the forum have tackled this before with reasonable success. Are you working with this kit too, or a similar kit from a different manufacture?

@apc

To overcome this problem I have used my own python scripts to split them by length, but this is not the best practice since reads length is a random measure. (emphasis mine)

This solution could be pretty good! When you plot the distribution of read lengths, do they truly look random, or do you see a bimodal distribution with peaks around the V3 and V4 lengths? If so, I think it would be defensible to split your reads based on these lengths, especially if you had paired end reads that could be joined to prevent overlap between regions.

16S ------V3====V3--V4=====V4-----
R1        >----------> overlapping regions :-(
R2    <----------<
joined    >------< just V3 :-)

---

@apc

A better solution is to align the reads using an appropriate reference and assign an H-region for each read.

I agree , but I don't know of a Qiime2-plugin that does this 'out of the box'

Colin