Can't run q2-SCNIC


(Elizabeth S Bent) #1

Hi, what version of qiime are you using to get the plugin to work?

It doesn’t install in 2018.11, I can install if but it won’t make correlations in 2018.6, and no matter what I do I can’t get the standalone SCNIC program to work with the sparcc correlation method (it works okay with pearson), except on one larger file. I can’t figure out from the documentation if this is an issue with the smaller single-treatment input files being too small, or if it’s a different issue. The error message I keep getting with the standalone program is:

Input triggered condition to perform clr correlation, this is not yet implemented

When I use files made with qiime2-2018.6, I can get the files filtered, but when I try to make correlations I get this error:

Plugin error from SCNIC:

File b’/tmp/fastsparbrxo8iia/correl_table.tsv’ does not exist

Debug info has been saved to /tmp/qiime2-q2cli-err-wwrhezd_.log

Help!


q2-SCNIC: A tool for making correlation networks, finding modules of observations and summarizing them
(Matthew Ryan Dillon) #2

(Michael Shaffer) #3

Hello @ebent,

There are currently versions of q2-SCNIC for 2018.6 and 2018.8 but the recommended release is for 2018.8. There will be a version for 2018.11 coming out in the next week or so.

That error usually indicates that the data set is not large enough for sparCC to calculated reliable correlations. How many samples do have after filtering?

After filtering how big are your data tables? How many features and how many samples?


(Elizabeth S Bent) #4
  1. Re: size of data set: you are right. I did some tests with data sets of different sizes. Larger ones (20 or more samples) worked fine, smaller ones (8 or 4 samples) only would work when I set -m to pearson.

  2. I don’t have my data handy but the files that were used with qiime2018.6 were, again small ones- plenty of features but only 4 samples.

This is all right, since correlations made with only a few samples are probably not very useful anyway, but the way the experiment was set up, combinations of treatments only have four replicates and it would have been nice to be able to compare these.


(Michael Shaffer) #5