gneiss Proportion plot explaination

Hello Guys
I ran qiime gneiss balance-taxonomy, using the output from Geniss as input. Nothing went wrong but I can’t really find inforamtion about the lighter blue area and darker blue area in the proportion plot. I wonder if anybody know what it corresponds to.
My guess is the light/dark blue area correspond to the denominator/numerator, respectively, of the y level selected in parameter.
Plus,I also would like to know what is the proportion on the x-axis.

thanks in advance

That’s a good looking heatmap.

The taxonomy plots look concerning - could you post the files used to generate that?

Hi, thanks for your reply. Files seen attached. I a performed at class level since when I went lower, the taxonomic information was not available and I think no matter which level it is performed at, the same feature in the proportion plot was reported.

y1L2_taxa_summary-gradient-hierarchy_19-272-Sewintod.qzv (374.6 KB)

Hi @Charly_CHEN, I meant the raw biom tables + tree.
I’m wondering if those Clostridia taxa are all zero, which is why they aren’t showing up in the proportion plots.

Hi, sorry for the late reply, maybe different time zone. Files are attached. I ran with Qiime2 2019.4 with following scripts.
qiime gneiss balance-taxonomy
–i-table Blank-Mock-filtered-table.qza
–i-tree gradient-hierarchy_19-272-Sewintod.qza
–i-taxonomy taxonomy_deblur_19-272-Sewintod.qza
–p-taxa-level 2
–p-balance-name ‘y1’
–m-metadata-file meta-Sewintod.txt
–m-metadata-column Contamination-Conc-num
–o-visualization y1L2_taxa_summary-gradient-hierarchy_19-272-Sewintod.qzv

Blank-Mock-filtered-table.qza (587.6 KB) gradient-hierarchy_19-272-Sewintod.qza (374.9 KB)
taxonomy_deblur_19-272-Sewintod.qza (355.0 KB) meta-Sewintod.txt (8.3 KB)

Regarding how to pull out the taxa of interest, here is the python code to do so

import qiime2
from skbio import TreeNode
from gneiss.util import NUMERATOR, DENOMINATOR
art = qiime2.Artifact.load('gradient-hierarchy_19-272-Sewintod.qza')
tree = art.view(TreeNode)
y1 = tree.find('y1')
num = y1.children[NUMERATOR].tips()
denom = y1.children[DENOMINATOR].tips()
num_names = [x.name for x in num]
denom_names = [x.name for x in denom]
pd.Series(num_names).to_csv('numerator.csv')
pd.Series(num_names).to_csv('denominator.csv')

Not exactly sure what is going on with the proportion_plot - it looks like a visualization bug.
Note that we have deprecated this function this release, and have been recommending users to look into aldex2 / songbird / ancom for differential abundance analysis in the meantime.

1 Like

Thanks for your reply, I did realize that new workflow was generated for differential abundance analysis.
But I would like ask about how to interpret proportion plot in general. What does the light blue and dark blue color block mean? Does it correspond to the numerator or denominator of the selected y level? Also, how was the proportion under ther graph of proportion plot generatred, waht does it mean? Thansk in advance.

Correct. The top (light blue) are supposed to represent numerator taxa and the bottom (dark blue) are supposed to represent denominator taxa. They are supposed to be displayed based on how differential they are across groups. The columns are supposed to represent the different sample groups.
Proportions are calculated by normalizing to 1 within a sample.

Thank you for your reply and patience!

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