How to interpret the analysis results from ANCOM ?

[yoshikisuzu (YOSHIKI)]

Hi,

I'm doing a study to investigate at the relationship between immune checkpoint inhibitors and the gut flora in patients with lung cancer.

I divided the patients into three group,PD(progressive disease),SD(stable disease),PR(partial response).

Then,I runned differential abundance tests using the ANCOM plugin.
The results are as follows.

The top two features appear to be abundant in the PD group.
I would describe it as statistically numerous. For example p-value,q-value,Kruskal-Wallis and so on.

If you have the best statistical means, please let me know.
Thank you very much.

yoshiki

[Mehrbod_Estaki]

Hi @yoshikisuzu,

I may need you to clarify what exactly you are looking for here, it is not very obvious to me.

@yoshikisuzu

The top two features appear to be abundant in the PD group.

Correct. ANCOM has identified these 3 features as important (and statistically different) across groups. From the table it is easy to see that the top 2 features are only present in PD and not at all in the other groups (remember that the 1.00 values you see there are psudeocounts that we add for ANCOM, and so the real value should be 0.00)
Feature C is only found in your SD group but not the other, and this was considered to be statistically significant by ANCOM.

@yoshikisuzu

I would describe it as statistically numerous.

I'm not sure what you mean by this.

@yoshikisuzu

For example p-value,q-value,Kruskal-Wallis and so on.

You do not get a p-value with ANCOM. Instead it relies on the W scores (Y axis) which represents the strength of your test, and the F score (X axis) which is a measure of the effect size. You can read more details about the results on this post here, and I also recommend reading the original ANCOM paper and ANCOM-II for further insight.