Producing a list differentially abundant OTU

(Yannick Gansemans) #1

Hi,

Is there a way to produce a list of differentially abundant features whzen comparing 2 conditions? Qiime1 seems to have differential_abundance.py to do something similar. I am a qiime 2 novice, so any assistance is greatly appreciated.

Yannick

Is there an equivalent of differential_abundance.py in qiime2?
(Colin Brislawn) #2

Hello Yannick,

Yep! You could try one of these plugins:
https://docs.qiime2.org/2018.11/plugins/available/composition/ancom/
https://docs.qiime2.org/2018.11/plugins/available/longitudinal/pairwise-differences/

By the way, you can find a full list of qiime 2 plugins on this page:
https://docs.qiime2.org/2018.11/plugins/available/

Colin

1 Like
(Yannick Gansemans) #3

Hi,

I am a qiime 2 novice and would like to produce a list of OTU that are differentially abundant between the two conditions I am comparing. For each of those differentially abundant features, would like to have an estimate of fold-change with a p value. Qiime1 seems (I have not tried it, done everything so far in qiime 2) to have a specific command for that: differential_abundance.py. How do I produce this kind of output in qiime2? Any help greatly appreciated.

Yannick

(Colin Brislawn) #4

Either one of those plugins should be able to do this.

Have you tried using one of those plugins? What command did you run and what output did it give you?

Colin

(Yannick Gansemans) #5

Dear Colin,

Thank you for the suggestions. I did try ANCOM before I posted my question. Unfortunately this plugin sometimes produces a large list of differential features where most have W=0. This cannot be correct?

Yannick

(Matthew Ryan Dillon) assigned colinbrislawn #6
(Colin Brislawn) #7

Would you feel comfortable sharing your ancom results? We can take a look at the results and see if they make sense.

In the ANCOM tutorial, many of the taxa have low w scores, so this might be similar to your data.

Colin

1 Like
(Matthew Ryan Dillon) unassigned colinbrislawn #8
(Yannick Gansemans) #9

Hi Colin,

Sorry for the late reply, I was out of office for a week. I am attaching the volcano plot and the two tables related to the ANCOM analysis. I also include a PCoA plot (unweighted unifrac) showing 2 distinct populations, so we really expect differential features. As explained, the ANCOM analysis does seem to list all of my OTUs, but only a handful appear to have sensible W values and too many just have W=0 and should not be in the table? When I collapse taxa to level 7 (species) or 6 (family), nothing is significant anymore.

Thanks again for your assistance.

ancom_results.zip (282.9 KB)

Yannick

(Mehrbod Estaki) #11

Hi @Yannick_Gansemans,
Have a look through this post which may shed some light on your low/zero W scores showing as significant, which ultimately are not to be trusted.

2 Likes
(Yannick Gansemans) #12

Thanks! I will give the ANCOM2 R code a try.

1 Like
(system) closed #13

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