coVarTypeStat: automatic covariate (y-variable) type based univariate...

In WMBEdmands/MetMSLine: an automated and fully integrated pipeline for rapid processing of high-resolution LC-MS metabolomic datasets.

Description

function determines whether a co-variate is two sample, continuous or up to n distinct classes (three or more sample) and selects the statistical method to apply accordingly.

Usage

coVarTypeStat(peakTable = NULL, obsNames = NULL, coVariate = NULL,
  continuous = T, nMaxClasses = 3, minSampPerClass = 3, Logged = T,
  base = 2, MTC = "none")

Arguments

peakTable	either a data.frame, full file path as a character string to a .csv file of a peak table in the form observation (samples) in columns and variables (Mass spectral signals) in rows. If argument is not supplied a GUI file selection window will open and a .csv file can be selected.
obsNames	character vector of observation (i.e. sample/ QC/ Blank) names to identify appropriate observation (sample) columns.
coVariate	numeric, character or factor vector of covariates (y-variables).
continuous	default = TRUE, if TRUE continuous variables are treated as such and correlation analysis is selected (spearman or pearson is selected based on whether the data is log-transformed or not see Logged argument). if FALSE continuous variables are split into a two sample variable.
minSampPerClass	default =3, minimum number of samples for a unique class else multiple class samples will be considered as continuous variables. (N.B. at least 5 for Mann-Whitney-U)
Logged	default = TRUE, has the data already been log transformed, if TRUE parametric methods will be selected (i.e. t-test (two sample), ANOVA (three or more sample), pearson correlation) if FALSE then non-parametric methods will be selected (i.e. t-test (two sample), ANOVA (three or more sample), pearson correlation). NB. if residuals from batchAdj contain negative values then the negatives will be shifted to the right with a constant value with the smallest lowest value equal to one prior to fold change calculation.
base	numeric this will be used to exponentiate log transformed data for mean/ median fold change calculation.
MTC	Multiple Testing Correction default is "none", see p.adjust.methods for details of options. ("holm", "hochberg", "hommel", "bonferroni", "BH", "BY", "fdr", "none")

Value

a named list (according to stat method selected) containing the 5 followning named elements :

varType = type of variable identified.

method = statistical method applied (either parametric or non-parametric according to whether data log transformed).

MTC = multiple testing correction method applied

coVariate = the co-variate supplied, if continuous is False the high/low two sample variable created for any continuous variables will appear here.

result = depending on type of variable supplied the result element will contain the following:

1. "Continuous" 4 column data frame of:

i. adjusted/ multiple testing corrected p-/q-values (dependent on MTC method)

ii. raw p.value (from cor.test function)

iii. estimated correlation coefficient (from cor.test function)

iv. mean or median fold change value (dependent on whether parametric or non-parametric method used). In the case of continuous variables a two sample high-/low- variable is created (using the cut function)

2. "two sample" 3 column data frame of:

i. adjusted/ multiple testing corrected p-/q-values (dependent on MTC method)

ii. raw p.value (from either t.test or wilcox.test function)

iii. mean or median fold change value (dependent on whether parametric or non-parametric method used).

3. "three or more sample" variable column (dependent on number of comparisons) data frame of:

i. adjusted/ multiple testing corrected p-/q-values (dependent on MTC method).

ii. raw p.value (from either ANOVA or wilcox.test function).

iii - number pairwise class comparisons. mean or median fold change columns. All pairwise class comparisons of fold change are carried out. (dependent on whether parametric or non-parametric method used).

raw p.values/ multiple testing corrected p-/q-values (dependent on method) and fold changes (if applicable).

A class must contain a minimum number of samples to be considered. if continuous equals true continuous variables

WMBEdmands/MetMSLine documentation built on May 9, 2019, 10:03 p.m.