aldex: Compute an 'aldex' Object
In ggloor/ALDEx_bioc: Analysis Of Differential Abundance Taking Sample and Scale Variation Into Account

aldex

R Documentation

Compute an `aldex` Object

Description

Welcome to the ALDEx2 package!

The aldex function is a wrapper that performs log-ratio transformation and statistical testing in a single line of code. Specifically, this function: (a) generates Monte Carlo samples of the Dirichlet distribution for each sample, (b) converts each instance using a log-ratio transform, then (c) returns test results for two sample (Welch's t, Wilcoxon) or multi-sample (glm, Kruskal-Wallace) tests. This function also estimates effect size for two sample analyses.

Usage

aldex(
  reads,
  conditions,
  mc.samples = 128,
  test = "t",
  effect = TRUE,
  CI = FALSE,
  include.sample.summary = FALSE,
  verbose = FALSE,
  paired.test = FALSE,
  denom = "all",
  iterate = FALSE,
  gamma = NULL,
  ...
)

Arguments

`reads`	A non-negative, integer-only `data.frame` or `matrix` with unique names for all rows and columns. Rows should contain genes and columns should contain sequencing read counts (i.e., sample vectors). Rows with 0 reads in each sample are deleted prior to analysis.
`conditions`	A character vector. A description of the data structure used for testing. Typically, a vector of group labels. For `aldex.glm`, use a `model.matrix`.
`mc.samples`	An integer. The number of Monte Carlo samples to use when estimating the underlying distributions. Since we are estimating central tendencies, 128 is usually sufficient.
`test`	A character string. Indicates which tests to perform. "t" runs Welch's t and Wilcoxon tests. "kw" runs Kruskal-Wallace and glm tests. "glm" runs a generalized linear model using a `model.matrix`. "corr" runs a correlation test using `cor.test`.
`effect`	A boolean. Toggles whether to calculate abundances and effect sizes.
`CI`	A boolean. Toggles whether to calculate effect size confidence intervals Applies to `test = "t"` and `test = "iterative"`.
`include.sample.summary`	A boolean. Toggles whether to include median clr values for each sample. Applies to `effect = TRUE`.
`verbose`	A boolean. Toggles whether to print diagnostic information while running. Useful for debugging errors on large datasets. Applies to `effect = TRUE`.
`paired.test`	A boolean. Toggles whether to do paired-sample tests. Applies to `effect = TRUE` and `test = "t"`.
`denom`	A character string. Indicates which features to retain as the denominator for the Geometric Mean calculation. Using "iqlr" accounts for data with systematic variation and centers the features on the set features that have variance that is between the lower and upper quartile of variance. Using "zero" is a more extreme case where there are many non-zero features in one condition but many zeros in another. In this case the geometric mean of each group is calculated using the set of per-group non-zero features.
`iterate`	A boolean. Toggles whether to iteratively perform a test. For example, this will use the results from an initial "t" routine to seed the reference (i.e., denominator of Geometric Mean calculation) for a second "t" routine.
`gamma`	A numeric. The standard deviation on the within sample variation.
`...`	Arguments to embedded method (e.g., `glm` or `cor.test`).

Details

See "Examples" below for a description of the sample input.

Value

Returns a number of values that depends on the set of options. See the return values of aldex.ttest, aldex.kw, aldex.glm, and aldex.effect for explanations and examples.

Author(s)

Greg Gloor, Andrew Fernandes, and Matt Links contributed to the original package. Thom Quinn added the "glm" test method, the "corr" test method, and the "iterate" procedure. Michelle Pistner Nixon and Justin Silverman contributed the scale and PPP routines

References

Please use the citation given by citation(package="ALDEx2").

Examples

# The 'reads' data.frame should have row
# and column names that are unique, and
# looks like the following:
#
#              T1a T1b  T2  T3  N1  N2  Nx
#   Gene_00001   0   0   2   0   0   1   0
#   Gene_00002  20   8  12   5  19  26  14
#   Gene_00003   3   0   2   0   0   0   1
#   Gene_00004  75  84 241 149 271 257 188
#   Gene_00005  10  16   4   0   4  10  10
#   Gene_00006 129 126 451 223 243 149 209
#       ... many more rows ...

data(selex)
selex <- selex[1201:1600,] # subset for efficiency
conds <- c(rep("NS", 7), rep("S", 7))
x <- aldex(selex, conds, mc.samples=2, denom="all",
           test="t", effect=TRUE, paired.test=FALSE)

ggloor/ALDEx_bioc documentation built on Oct. 31, 2023, 1:13 a.m.