Hopefully a quick question/confirmation. I am running qiime diversity alpha-correlation on a numerical metadata variable (age) and Faith’s phylogenetic diversity. My result returns a p-value of “0.0000” for both Pearson and Spearman methods. Is this normal/expected? Does it just mean the correlation is very very unlikely due to chance?
Here is my command: qiime diversity alpha-correlation --i-alpha-diversity Age_diversity_30000/faith_pd_vector.qza --m-metadata-file ../ProjectName_2015_map_MHD_20190903_nocontrols_noMissingAge.txt --o-visualization Age_diversity_30000/faith_pd_correlation.qzv
It might be a floating point error, because it’s possible to get a zero p-value in python when you exceed what they can process. (I get them sometimes on chi-squared tests where the distribution kinda groans and goes, “your test statistic is 2500, yes, it’s significant! ”). So, i would look at your correlation coeffecient and if thats close to 1, your p-value may make more sense.
Although I might also double check your metadata just to be sure because it seems strong for microbiome data. I have a good day when my R^{2} ≥ 0.1 with alpha diversity.
Thanks for your advice! The correlation coefficient isn’t super high, it’s about 0.3-0.4 (which I agree is still high for microbiome data). Since I’m looking at human oral microbiomes and the age range covers events like transition from baby to adult teeth and increased exposure to oral disease risk factors over time, that seems high but not outlandishly so to me …
”). So, i would look at your correlation coeffecient and if thats close to 1, your p-value may make more sense.