Analysis of UBiome Data - Roadblocks & Questions

To begin, I should mention I have previous experience with De Novo RNA Seq but am relatively new to QIIME, so forgive my ignorance.

In short, I jumped onboard with a new lab and we are attempting to analyze human gut microbiota using Illumina data generated by the UBiome company. Amplicon 515f-806r length 390. The data that was supplied to us has already been demultiplexed, and has had the Linker and Primer Sequences removed. The sequences themselves have been provided to us, but they are absent from the reads themselves.

Now, assuming that’s not an issue, I began by using just one Subject’s data to setup a protocol for use with the other subjects. Namely, operating within a folder containing Sample_R1_L001.fastq, Sample_R1_L002, Sample_R2_L001.fastq, and Sample_R2_L002. Each file is about 30MB.

First, I attempted to process paired-end reads with QIIME2 using
However, the assembled reads generated an unreasonably small output file (<50kb).
After searching forums, I learned it might be better to forego merging and just use Join_Paired_Ends on R1/R2 from Lane001, which didn’t really improve the pairing output file size. I did increase the --perf_max_diff, which increased the output file size, but only when I increased it to about 20% (which I assume is just not great for the validity of my analysis, given it’s accepting a bunch of random base pairs at that point).

I should mention that when I began, I didn’t notice how small the merged file was and was able to quality filter, do OTU picking, and generate a Rank Abundance Graph for it, but it contained very little data so I wouldn’t consider it very usable.

Any suggestions on how I can rectify these problems and move forward with this project? I’m very motivated and happy to provide whatever information is necessary. Thank you for your time.



Wonderful! That should make analysis very straight-forward.

QIIME 2 doesn’t have scripts like QIIME 1 did. Instead we have commands such as qiime vsearch join-pairs. These are provided by plugins which give methods and visualizers.

Since you are starting out, this is a really great time to learn QIIME 2. It’s very different from QIIME 1, but we think it’s a bit easier to start out with. I would recommend starting with our documentation. Of particular note is the moving pictures tutorial and the importing tutorial. And of course please use the forum as a resource, we’re all here to help!

As it relates to your original question, I think you’ll find the modern denoising methods like DADA2/Deblur will give you better results (and will probably come with their own set of questions :slight_smile: ).

Thank you for the prompt reply, I will look into DADA2/Deblur and the linked documentation

1 Like

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.