# p-value for taxonomic classification?
**URL:** https://forum.qiime2.org/t/p-value-for-taxonomic-classification/9758
**Category:** General Discussion
**Tags:** composition, ancom
**Created:** [May 14, 2019, 7:15am UTC](https://forum.qiime2.org/t/p-value-for-taxonomic-classification/9758 "2019-05-14T07:15:59Z")
**Posts on this page:** 2
**Page:** 1
### Author:  [@Shruthi](https://forum.qiime2.org/u/Shruthi)
#### Post date: [May 14, 2019, 7:15am UTC](https://forum.qiime2.org/t/p-value-for-taxonomic-classification/9758/1 "2019-05-14T07:15:59Z")
Hi,
I was wondering how to estimate the p-value for statistical comparison between the phyla of two different subjects. Is there a way to do this in QIIME 2?
For example, I require something like this " The phyla abundance composition differed significantly between intestinal and faecal samples (PERMANOVA: _P_ = 0.002, pseudo r2 = 0.09; Permutest: _P_ = 0.093). Intestinal samples contained an enrichment of _Proteobacteria_ compared with faecal samples (15.71-fold change, _P_ \< 0.001) and more _Actinobacteria_ (2.09-fold change, _P_ = 0.002). Faecal samples contained significantly more _Lentisphaerae_ (1.74-fold change, _P_ = 0.050)."
Thanks in advance.
---
### Author:  [@jwdebelius](https://forum.qiime2.org/u/jwdebelius)
#### Post date: [May 14, 2019, 11:16am UTC](https://forum.qiime2.org/t/p-value-for-taxonomic-classification/9758/3 "2019-05-14T11:16:52Z")
Hi @Shruthi,
Depends on what you want to use to do your analysis, but the short answer is not the way you'd like.
In QIIME, the primary implementation is ANCOM, which calculates a W statistic. Please check out this prior discussion around why the ANCOM W is not a P and ways to approach that.
> [@How to calculate adjusted P-value for a null hypothesis](https://forum.qiime2.org/t/how-to-calculate-adjusted-p-value-for-a-null-hypothesis/7098):
>
> Hello, I examine the significance of differences in taxa composition between two groups using the ANCOM package. I would like to make a bar plot for p-values calculated by ANCOM. I know that the current ANCOM does not provide p-values, so I am wondering if I could use the mean of p-values across sub-nullhypotheses as well as calculating the W value by counting the number of rejecting sub-nullhypotheses. Is it reasonable? Could anyone give me suggestions?
An alternative is to use `q2-gneiss`, PhILR or Phylofactor (both in R) to get a split based on a partition in the data. But, that won't give you exactly what you want, either.
Best,
Justine