Plugin error from mmvec: index 0 is out of bounds for axis 0 with size 0

Community Plugin Support
1

Hello everyone,

I have been trying to create a heatmap with mmvec and I keep getting this error:

Plugin error from mmvec:
index 0 is out of bounds for axis 0 with size 0
Debug info has been saved to /tmp/qiime2-q2cli-err-j1eob9aj.log

I have seen a similar post with the same error but I don't think mine is caused by the same issue that @jhaffner09 was facing.

I have attached here my conditional.txt that was imported in qiime2 successfully. I have also attached my metabolite metadata and microbe metadata.

I have run this code

qiime mmvec heatmap 
--i-ranks ranks.qza
--m-microbe-metadata-file taxonomy_fourth_trial.txt
--m-microbe-metadata-column taxon
--m-metabolite-metadata-file metabolite_metadata_fourth_trial.txt
--m-metabolite-metadata-column Compound_name
--p-level -1
--o-visualization ranks_heatmap.qzv
--verbose

And this what I got as a result

image
image707×663 22.1 KB

conditionals_modified.txt (11.7 KB)
metabolite_metadata_fourth_trial.txt (1.1 KB)
taxonomy_fourth_trial.txt (18.5 KB)

Any help would be very much appreciate it!
Thank you very much,

Danisa

4

Hi @DanisaBescucci , it looks like you are generating a unique color for every taxonomic species -- if I had to take a guess there are not enough colors to uniquely color every taxa. What happens if you try running this at the class level instead?

CCing @Nicholas_Bokulich who helped design the mmvec heatmaps.

1 Like
5

Hi Jamie,

I tried with class level instead and I keep getting the same error. Could this be related with the file that I am using as rank.qza? I am not sure if I understood correctly how to create that file. Do you happen to have an example on how to go from the conditionals.qza that MMvec paired-omics gives as a result, and transform that into a rank?
I am sorry if this a very basic question :frowning: I am just trying to make sure that I am approaching this correctly.
Thank you in advance,
Danisa

6

Hi @DanisaBescucci , the conditionals.qza can be ranked (aka should be treated as the equivalent as rank.qza). I think this just a plotting issue. Unfortunately I don't have time at this very moment to carefully look into the visualization code -- so I've raised an issue on github for now.

But you can pull out a text file with the full table of conditional probabilities as follows

qiime tools export conditionals.qza

With this you can visualize this table with any program that you'd like. For instance, if you wanted to visualize a heatmap of the probabilities with seaborn / python it would look something like as follows

import seaborn as sns
import pandas as pd
conditionals = pd.read_table('conditionals/conditionals.tsv', index_col=0)
sns.clustermap(conditionals)

See seaborn clustermap for more information about customizable colors, etc. Jupyter notebook may also be worth looking into for quickly generating plots.

7

@mortonjt & @DanisaBescucci,

I am working on the heatmap code for the new mmvec release, I will take a look at this.