Comparing v3 and v4 datasets with v3-v4 amplicon datasets

Question

[bioinfo]

I have analysed 50 16S amplicon datasets using the v3-v4 regions (PE, 250 bp, merged, and ~450 bp merged reads formed) and generated an OTU table. Suddenly, we have recieved another 10 samples that were sequenced individually for v3 (PE, 300 bp) and v4 (PE, 300 bp) regions. Is there a way to merge/join these two individual datasets before calling OTUs for all 60 samples using v3-v4?

I am trying to avoid trimming v3 or v4 parts from the large datasets to make individual v3 and v4 datasets and then caling OTUs separately using v3 and v4 regions.

Any suggestions or ideas would be apprecited.

[h.mon]

There is a published method for combining different amplicon regions in the same analysis, you should check the paper and see if it applies to your case:

Combining 16S rRNA gene variable regions enables high-resolution microbial community profiling

I haven't used it, so I have no idea if it works, but the fact it is coded in Matlab already raises a warning signal.

Answer 1

[Asaf]

You have two issues here (at least) - one is how to align them and the other is to account for the amplification bias in each of the primer sets. Honestly, I don't see a way to merge the separate V3 and V4. If you'll chop the V3-V4 into separate V3 and V4 you will end up with low quality bases that were initially in the middle of the sequence being at the ends and might not align well with the separate V3 and V4. You can call OTUs from the tree sets (V3-V4, V3 and V4) separately and then combine the analysis on the genus level and above, making sure that the separate V3 and V4 give the same overall results and making sure to include the batch as a confounding effect in the linear model you'll be applying.

I'm sorry but I don't see an easy way out.

[bioinfo]

Thanks for sharing your opinion! Calling OTUs from three datasets separately would not be a bad idea to start with.