Are the log fold changes reported in the ancom-bc analysis 10-based or 2-based ?

I would assume 2-based. Check the Ancom docs and I would also check the numbers. (Travelling so I can't check myself.) We should have this in our docs so make an issue once you've found out.

confusing: from the original ancom-bc paper docs.....it seems to be natural log (ln) Value: • `feature.table`: Data frame or matrix. Return the input

feature.table

. • `res`: Data frame. The primary result of ANCOM-BC consisting of: ◦ `mean.difference`: Numeric. The estimated mean difference of absolute abundance between groups in log scale (natural log); ◦ `se`: Numeric. The standard error of `mean.difference`; ◦ `W`: Numeric.

mean.difference/se

, which is the test statistic of ANCOM-BC. ◦ `p.val`: Numeric. P-value obtained from two-sided Z-test using the test statistic

W

. ◦

q.val

. Numeric. Q-value obtained by applying

adj.method

to

p-val

. ◦

diff.abn

. Logical. TRUE if the taxon has

q.val

less than

alpha

. • `d`: A numeric vector. Estimated sampling fractions in log scale (natural log). • `mu`: A numeric vector. Estimated log (natural log) mean absolute abundance for each group. • `bias.em`: Numeric. Estimated mean difference of log (natural log) sampling fractions between groups through E-M algorithm. • `bias.wls`: Numeric. Estimated mean difference of log (natural log) sampling fractions between groups through weighted least squares. from the official documentation it just says log fold change. "Claude" found the answer (hopefully good): it's natural log since the code uses R's log() function which defaults to ln

Unusual, but since the only doc that specifies this and Claude agrees, I'm convinced. Make a PR updating the ampliseq docs please