ZicoSeq: Permutation-based differential abundance analysis In chloelulu/ZicoSeq: Has not been decided

Description

Permutation-based differential abundance analysis

Usage

 ``` 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34``` ```ZicoSeq( meta.dat, comm, grp.name, adj.name = NULL, prev.filter = 0.1, abund.filter = 10, min.prop = 0, is.winsor = TRUE, winsor.qt = 0.97, is.prior = TRUE, prior.dist = c("BetaMix", "ZIBB"), post.method = c("sample", "mean"), post.sample.no = 25, link.func = list(function(x) x^0.25, function(x) x^0.5, function(x) x^0.75), link.d.func = list(function(x) 0.25 * x^(-0.75), function(x) 0.5 * x^(-0.5), function(x) 0.75 * x^(-0.25)), variance.EB = FALSE, df.prior = 10, perm.no = 99, strata = NULL, stats.combine.func = max, stage.no = 6, topK = NULL, stage.fdr = 0.75, stage.max.pct = 0.5, is.fwer = FALSE, is.tree.fdr = FALSE, tree = NULL, verbose = TRUE, return.comm = FALSE, return.perm.F = FALSE, ... ) ```

Arguments

 `meta.dat` a data frame containing the sample information `comm` a matrix of counts, row - features (OTUs, genes, etc) , column - sample `grp.name` a character, variable of interest; it could be numeric or categorical; should be in "meta.dat" `adj.name` a character vector, variable(s) to be adjusted; they could be numeric or categorical; should be in "meta.dat" `prev.filter` features with prevalence (i.e., nonzero proportion) less than "prev.cutoff" or be filtered `abund.filter` features with a total counts less than "abund.cutoff" or be filtered `min.prop` Undetermined `is.winsor` a logical value indicating whether winsorization should be performed to replace outliers. The default is TRUE. `winsor.qt` the winsorization quantile, above which the counts will be replaced `is.prior` a logical value indicating whether to perform posterior inference based on some prior distribution on the proportion data `prior.dist` prior distribution, either two-component beta-binomial mixture ("BetaMix") or zeroinflated beta-binomial ("ZIBB") `post.method` method for posterior inference, either based on posterior sampling ("sample") or approximate posterior mean ("mean") `post.sample.no` the number of posterior samples if posterior sampling is used `link.func` a list of functions that connects the ratios to the covariates `link.d.func` a list of the derivative function of "link.func"; only need to specifiy when "post.method" is "mean" `variance.EB` a logical value indicating whehter to perform empirical Bayes based variance shrinkage `df.prior` the degree of freedom of the prior inverse gamma distribution for variance shrinkage `perm.no` the number of permutations; If the raw p values are of the major interest, set "perm.no" to at least 999 `strata` a factor indicating the permutation strata; permutation will be confined to each stratum `stats.combine.func` function to combine the F-statistic for the omnibus test `stage.no` the number of stages if multiple-stage ratio stategy is used `topK` the number of dominant features that will be excluded in the initial stage ratio calculation `stage.fdr` the fdr cutoff below which the features will be excluded for calculating the ratio `stage.max.pct` the maximum percentage of features that will be excluded `is.fwer` a logical value indicating whether the family-wise error rate control (West-Young) should be performed `is.tree.fdr` a logical value indicating whether tree-based false discovery rate shuold be carried out `tree` a class of "phylo", the tree relats all the OTUs, and should have the same names in "comm" `verbose` a logical value indicating whether the trace information should be printed out `return.comm` a logical value indicating whether the wisorized, filtered "comm" matrix should be returned `return.perm.F` Undetermined `...` arguments passing to tree-based fdr control

Value

A list with the elements

 `call` the call `comm` the wisorized, filtered "comm" matrix `filter.ind` a vector of logical values indicating which features are tested `R2` a matrix of percent explained variance (number of features by number of functions) `F0` a matrix of F-statistics (number of features by number of functions) `RSS` a matrix of residual sum squares (number of features by number of functions) `df.model, df.residual` degree of freedoms for the model and residual space `p.raw` the raw p-values based on permutations (not accurate if "perm.no" is small) `p.adj.fdr` permutation-based FDR-adjusted p-values `p.adj.tree.fdr` permutation-based tree FDR-adjusted p-values `p.adj.fwer` permutation-based FWER-adjusted (West-Young) p-values `tree.fdr.obj` the object returned by the "TreeFDR"

chloelulu/ZicoSeq documentation built on Nov. 4, 2019, 8:50 a.m.