# ANCOM result volcano plot clr F-statistic calculation formula
**URL:** https://forum.qiime2.org/t/ancom-result-volcano-plot-clr-f-statistic-calculation-formula/9168
**Category:** General Discussion
**Tags:** ancom
**Created:** [April 8, 2019, 2:15pm UTC](https://forum.qiime2.org/t/ancom-result-volcano-plot-clr-f-statistic-calculation-formula/9168 "2019-04-08T14:15:06Z")
**Posts on this page:** 3
**Page:** 1
### Author:  [@Josh\_Hyungyum\_Kim](https://forum.qiime2.org/u/Josh_Hyungyum_Kim)
#### Post date: [April 8, 2019, 2:15pm UTC](https://forum.qiime2.org/t/ancom-result-volcano-plot-clr-f-statistic-calculation-formula/9168/1 "2019-04-08T14:15:07Z")
Hi,
I'm conducting differential abundance testing with ANCOM and have a question with the clr F-statistic calculation. I've checked the forum and found good explanations and intuitions regarding [negative F-statistic with only 2 groups](https://forum.qiime2.org/t/negative-f-statistic-values-in-ancom-output-after-filtering-low-abundant-taxa/3211), [effect size perspective of X-axis](https://forum.qiime2.org/t/how-to-interpret-ancom-results/1958/2), [percentile abundance interpretation](https://forum.qiime2.org/t/interpreting-values-from-an-ancom-percentile-abundance-table/1497), and [W-statistic hypothese testing](https://forum.qiime2.org/t/specify-w-cutoff-for-anacom/1844/9).
For me, however, the formula of clr F-statistic calculation (more than 2 groups case) is still sketch even though I somehow understand it's meaning and intuition. Since it's not explicitly mentioned in original ANCOM paper and its supplemental material, I hope one of the devs here could provide it's implementation to qiime2.
Could anyone explain this in terms of F-test with good annotations like @mortonjt did in [here](https://forum.qiime2.org/t/specify-w-cutoff-for-anacom/1844/9)?
Thanks in advance,
Josh
---
### Author:  [@mortonjt](https://forum.qiime2.org/u/mortonjt)
#### Post date: [April 8, 2019, 8:02pm UTC](https://forum.qiime2.org/t/ancom-result-volcano-plot-clr-f-statistic-calculation-formula/9168/2 "2019-04-08T20:02:30Z")
We wanted to provide a way to visualize something that looks like a volcano plot with multiple categories.
If you have just 2 categories, easy -- you just take the difference.
If you have more than 2 categories, then what would you do?
The easiest approach would be to compute something that looks like the F-statistic as a proxy for the differences across groups.
---
### Author:  [@system](https://forum.qiime2.org/u/system)
#### Post date: [May 10, 2019, 2:02am UTC](https://forum.qiime2.org/t/ancom-result-volcano-plot-clr-f-statistic-calculation-formula/9168/3 "2019-05-10T02:02:35Z")
This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.