Need Visualization tools for Nanopore microbiome NGS data


(Shashi Kanth) #1

Hi,
i have tried using Qiime2 tool for visualization.i have gone through the tutorial in the website but couldn’t find appropriate results for visualization. i have two barcode fastq files generated from Nanopore NGS Sequencing i have tried Qiime import and export , extract but couldn’t visualize the data if anyone please help me with this.
is there any alternative visualization tool for microbiome data

Thank you


(Devon O'rourke) #2

Hi @Shashi_Kanth,
I’ve worked with enough Nanopore data to have a good feeling for some of the tools out there. I’m curious what steps you took prior to importing your data to QIIME?
What are you trying to visualize? A barplot with taxonomic information? A box and whiskers for some diversity metric? Something PCoA style?
There are many angles you can explore data visually within QIIME - what did you have in mind specifically? You may also think about using external packages too; something like phyloseq for a few other ideas related to exploring the data. If you’re looking for metrics related to the Nanopore run itself - things like distributions of quality scores before and after trimming, for instance, you should look into a wonderful set of tools called Nanoplot by Wouter DeCoster. There’s CLI and GUI interface for that package.
Cheers,


(Shashi Kanth) #3

Hi @devonorourke,
Thanks for the reply.
yes im looking for visualisation of taxonomic information and other stuff. i have gone through the plugins available in QIIME.
i have tried analyzing my data following the tutorials https://docs.qiime2.org/2018.11/tutorials/qiime2-for-experienced-microbiome-researchers/#data-processing-steps
https://docs.qiime2.org/2018.6/tutorials/importing/#fastq-manifest-formats

The nanopore data which we have is barcode3.fastq (23mb) and barcode6.fastq (60MB) files when i tried using the command it shows error.

I tried to import my data using the manifest but shows an error. the screenshots was attached

.

can you please help me out with QIIME tool.

whether QIIME is the got tool to work with Nanopore dataset.

i got results with NanoPlot, NanoStats toolkits.

Thank you


(Devon O'rourke) #4

Hey @Shashi_Kanth,
You mention you have 2 files - barcode3 and barcode6. From what I can tell, each of those files are in a folder you’ve called fastq_files, correct?
When I’ve imported my data (Nanopore derived or otherwise), I’ve used the manifest format approach you’ve linked above, and like what you’ve attempted to do in the second of two screen shots.
I noticed you’re passing the name of the directory (fastq_files) for the input argument, and I suspect that’s where your error is coming from.
Here’s the silly/confusing thing: when you try to follow their directions in the import tutorial using an EMP-format what you are supposed to use as the --input-path parameter is the folder name (which is what you did in your first screenshot). However, when you follow their tutorial for manifest style imports, the same --input-path parameter is not the folder name, rather, it’s the manifest file itself.
Assuming you’ve created a manifest file following their guidelines, you just need to switch up one value in your second screenshot: substitute fastq_files for whatever the name is for your manifest file (if it’s not in the same directory as where you are executing the command, make sure it’s the full path or use an environmental variable set to that path).
That should (hopefully) do the trick.

One follow up question for you and one other comment:
Question: did you filter these reads already? You absolutely should have already corrected these reads with Nanopolish prior to trying to make some sort of taxonomic classification… these reads are far noisier than standard Illumina ones that the denoising algorithms defaults are used to… though note that DADA2 is now making headway into correcting long reads (though these are PacBio data thus far), and I’m not sure if QIIME’s current DADA2 plugin can do anything reasonable with Nanopore reads. That’s on my fall 2019 to do list, so I can’t help you there yet. Maybe @benjjneb can speak to what DADA2 can do in QIIME with long, noisy reads.

Comment: You don’t need QIIME to do any of this. I love QIIME and continue to find more ways to use it’s tools in my projects, but you may want to look at other tools like Centrifuge to get started with packages built for Nanopore data primarily (when it comes to taxonomic assignment).

Good luck


(Ben Callahan) #5

Nanopore data is still too noisy to identify exact sequences (ASVs) from, you’ll need a different solution for that kind of data. PacBio CCS data can be processed into ASVs with the latest DADA2 R package, and that feature will come to the Q2 once the new versions propagate through to the plugin.


(Devon O'rourke) #6

Is that true for1d2 reads too @benjjneb?
Maybe the R10 flow cell will help make things less noisy…
Thanks for the info!


(Ben Callahan) #7

Yes also for 1d2. The new flow cell will improve things, but it still won’t allow ASV level precision. It’s not knock on Nanopore tech, you can’t get ASVs from raw PacBio data either, but with PacBio you can construct CCS reads that have very high per-base accuracy (~99.9%).

Long-term I think that Nanopore is looking at multi-pore tech that could give an OOM increase in their per-base accuracy which would be very interesting and open up some new high-resolution applications.