NEWS
DiPPER 0.99.1
data.type is now a required argument with no default. A new value,
"abundance", covers non-count, non-proportional abundance data
such as CPM or TPM values.read.depth must now be chosen explicitly for count data. For other data
types, it defaults to FALSE. Moreover, read depths supplied as a
metadata column are now log10-transformed and added to the model
automatically.print() now reports also the estimated values of the hyperparameters of the
hierarchical Laplace prior. For this to work keep.pars of dipper() now
defaults to c("beta", "tau", "nu").verbose argument is added to dipper(). Setting verbose = FALSE
silences all messages, including those from Stan.threshold is now always interpreted on the scale of the supplied data.
For instance, for data.type = "counts", threshold = 2 means that count
values of 2 or below are interpreted as absence.- The vignette and the details section of
?dipper have been updated regarding
the type of the input data, controlling for sequencing depth, and the
choice between the symmetric and the asymmetric hierarchical prior.
- Bugs in parsing model formula have been fixed.
DiPPER 0.99.0
- Initial Bioconductor submission.
- Provides an implementation of DiPPER (Differential Prevalence via
Probabilistic Estimation in R), a Bayesian hierarchical modeling approach for
differential prevalence analysis, especially in microbiome studies.
- Supports both cross-sectional and longitudinal (repeated measures) designs,
with covariate adjustment and automatic control for sequencing depth.
- Posterior inference is performed with CmdStan, which must be installed
separately before installing DiPPER.