ANCOM and W-value

[Bruno_Andrade]

Dear Qiime2 users and developers, I am using ANCOM to perform Differential abundance analysis but I have some doubts regarding the w-value. After reading this excelent topic I ended up understanding what the w-value means but, after testing in some of my samples, I realized that every Differentially abundant ASV is true if the alternative sub-hypothesis (if the ASV is significantly different) to equal or more than 90% of the total ASVs in the sample. Is my observation true?

[Nicholas_Bokulich]

Hi @Bruno_Andrade,

@Bruno_Andrade

Differentially abundant ASV is true if the alternative sub-hypothesis (if the ASV is significantly different) to equal or more than 90% of the total ASVs in the sample.

See the notes about ANCOM's assumptions here.

ANCOM assumes that the minority of features (< ~25%) are differentially abundant. If you are comparing very different sample types, it will be prone to false positives. See the original paper for more details.