This model is applied on data of a single gene. It will infer effect size for both sample-level variable and positional level functional annotations. We used an EM algorithm to infer parameters.

ddmodel(mut, e, mr, fe, label, cov = NULL, ...)

Arguments

mut

a matrix of mutation status 0 or 1, rows positions, columns are samples.

e

a vector,phenotype of each sample, should match the columns of mut and mr

mr

a matrix, mutation rate of each sample at each mutation (log scale) that is not dependent on sample level factor

fe,

a vector, increased mutation rate at each position, depending on e (log scale), should match the rows of mut and mr

cov

an optional numeric matrix of sample-level covariates (e.g. age, sex), with rows matching the columns (samples) of mut. When supplied, the covariates are added to the logistic regression of the EM responsibilities under both the null and alternative models, so the phenotype effect is tested conditional on the covariates. Default NULL (no covariates).