aldex2 output differentials

leod · August 2, 2021, 9:59am

Dear all,

I ran aldex2 on my data and got some nice effect charts and tables. Unfortunately, I don't understand what the table is telling me. I understood that each red dot in my effect diagram shows a significantly different ASV (not in test1), but I can't link the table to it.... Can you help me?

Thank you!!
L

differentials-test1.tsv (1.3 KB) test1.qzv (379.2 KB) differentials-test2.tsv (5.0 KB) test2.qzv (396.2 KB)

leod · August 2, 2021, 10:27am

And can I see somewhere where my two tested variables are (y-axis effect plot)? Or is that only possible with R?

colinbrislawn · August 3, 2021, 1:21am

Hello @leod,

Welcome to the forums!

You probably found this already, but I wanted to make sure I provided the link to the ALDEX2 qiime2-plugin and to the part of the vignette that explains what all those outputs mean. (Those are the columns in your test1 and test2 tsv files!)

As mentioned in the vignette, the values of your Control and Riesling group are listed in that table, along with the mean difference in value, p-value, and corrected p-value for various stat tests on each feature.

Let me know if you have specific questions and I can tell you more!

Colin

leod · August 3, 2021, 9:17am

Hello Colin,

thank you very much!
It is getting a little bit clearer now... But when I look at the corrected p-values, only some of the features are significant, even though they are all shown as red dots in the effect plot? Is that correct?
It is not quite clear to me yet, how to read the values of my Control and Riesling group in the table (sorry!!). Is there an x and a y value for each feature? So that I can find it again in the effect plot?
I would like to label the y-axis as in the attached image in red. Is this even an option, or is it not possible to generate these statements with this table?

Thank you for your help!
Leo

effectplot_2|638x500

colinbrislawn · August 3, 2021, 4:31pm

Hello Leo,

The 21 red points in the plot you posted are the same 21 ASVs in differentials-test2.tsv

x = diff.win (Dispersion)
y = diff.btw (Differance)

The x and y axis of each plot match up to columns in that table.

Now it's your turn: can tell me what columns were used to make the other three plots?

leod · August 4, 2021, 11:48am

Hi Colin,

aaaah okey, I see! So it is not possible to split the plot in half like in the figure I attached before?

Hmm, I will try!

MA plot:
y = diff.btw
x = rab.all

Volcano plot:
x = diff.btw
y = we.eBH

Effect size vs. qscore
x = effect
y = we.eBH

Is that right?

Thank you very much!

colinbrislawn · August 4, 2021, 1:32pm

Correct! Well done!

I had to look up the code to make sure I was also getting it right. You can check my work:

github.com

ggloor/q2-aldex2/blob/14602edc86be83255629ac41f5693cadcc0c8822/q2_aldex2/_visualizer.py#L67


      
          ########################################################################
          # close current plot, start one for MA
          plt.close()
          
          # base effect plot to build on
          plt.scatter(x="rab.all", y="diff.btw", data=table, color="grey", s = 6)
          
          # colour for significant points
          plt.scatter(x="rab.all", y="diff.btw", data=table[called],
                      color="red", s = 6)
          
          # change titles and labels
          plt.suptitle('MA plot')
          plt.xlabel('Relative Abundance')
          plt.ylabel('Difference')
          
          img_ma = os.path.join(output_dir, "ma_plot.png")
          plt.savefig(img_ma)
          
          ########################################################################
          # close current plot, start one for volcano

github.com

ggloor/q2-aldex2/blob/14602edc86be83255629ac41f5693cadcc0c8822/q2_aldex2/_visualizer.py#L86


      
          ########################################################################
          # close current plot, start one for volcano
          plt.close()
          
          # base effect plot to build on
          plt.scatter(x="diff.btw", y="we.eBH", data=table, color="grey", s = 6)
          
          # colour for significant points
          plt.scatter(x="diff.btw", y="we.eBH", data=table[called],
                      color="red", s = 6)
          
          # change titles and labels
          plt.suptitle('Volcano plot')
          plt.xlabel('Difference')
          plt.ylabel('Q score')
          
          img_volcano = os.path.join(output_dir, "volcano_plot.png")
          
          # change p values to log scale
          plt.yscale('log')
          # get min and max to plot for p values

github.com

ggloor/q2-aldex2/blob/14602edc86be83255629ac41f5693cadcc0c8822/q2_aldex2/_visualizer.py#L119


      
          # close current plot, start one for effect vs q score
          plt.close()
          
          # base effect plot to build on
          plt.scatter(x="effect", y="we.eBH", data=table,
                      color="grey", s = 6)
          
          # colour for significant points
          plt.scatter(x="effect", y="we.eBH", data=table[called],
                      color="red", s = 6)
          
          # change titles and labels
          plt.suptitle('Effect size vs q score')
          plt.xlabel('Effect size')
          plt.ylabel('Q score')
          
          # change p values to log scale
          plt.yscale('log')
          # get min and max to plot for p values
          minimum = table["we.eBH"].min()
          maximum = table["we.eBH"].max()

Those plots are all separate .png files, so you can save them one at a time. Is that's what you mean?
You could also replicate these plots directly in Python or R if you need more control.

leod · August 6, 2021, 12:00pm

Hi Colin,
great, thank you for your help!

Thank you for the codes, I think I am getting it now.

system · September 6, 2021, 6:01pm

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