Implements the default scale model described in Nixon et al. (Beyond
Normalizations / scale-uncertainty framework). This model generalizes the
centered log-ratio (CLR) normalization by treating the (log) scale as a
latent random variable and allowing additive uncertainty around the
CLR-implied scale differences via a Gaussian term with standard deviation
gamma.
clr.sm(X, logComp, gamma = 0.5)A numeric matrix of dimension N x nsample giving Monte Carlo samples
of the log-scale for each sample (rows) and each Monte Carlo draw (columns).
A numeric design matrix used to model scale variation across
samples. This is the covariate/design matrix passed internally by
aldex() to the scale model. Rows correspond to regression
coefficients (e.g., intercept and covariates after contrasts/encoding) and
columns correspond to samples. If the analysis includes only an intercept
(no covariates), X is typically a 1 x N matrix of ones. (This
parameter is automatically passed by aldex)
A numeric array of Monte Carlo log-compositions with
dimensions features x samples x nsample. This is produced
internally by ALDEx3 from Dirichlet-multinomial Monte Carlo sampling and
log-ratio representation. ##' (This parameter is automatically passed by aldex)
Non-negative scalar. Standard deviation of the Gaussian
perturbation that relaxes the CLR assumption about scale. gamma = 0
yields the pure CLR assumption; recommended default values in the
scale-uncertainty literature are often around 0.5, but appropriate
values depend on how strongly you trust the CLR scale assumption in the
current study.
Justin Silverman
In the limit gamma = 0, this reduces to the CLR assumption (no scale
uncertainty beyond the CLR-implied scale). Larger gamma values
represent increasing uncertainty about the CLR-implied scale differences.
Nixon G, Gloor GB, Silverman JD (2025). "Incorporating scale uncertainty in microbiome and gene expression analysis as an extension of normalization". Genome Biology. tools:::Rd_expr_doi("10.1186/s13059-025-03609-3")