check.associations: Check and visualize associations between features and classes

In KonradZych/SIAMCAT: Statistical Inference of Associations between Microbial Communities And host phenoTypes

Description

This function calculates for each feature a pseudo-fold change (geometrical mean of the difference between quantiles) between the different classes found in labels.

Significance of the differences is computed for each feature using a Wilcoxon test followed by multiple hypothesis testing correction.

Additionally, the Area Under the Receiver Operating Characteristic Curve (AU-ROC) and a prevalence shift are computed for the features found to be associated with the two different classes at a user-specified significance level alpha.

Finally, the function produces a plot of the top max.show associated features, showing the distribution of the log10-transformed abundances for both classes, and user-selected panels for the effect (AU-ROC, Prevalence Shift, and Fold Change)

Usage

check.associations(siamcat,fn.plot,color.scheme = "RdYlBu",
    alpha =0.05,mult.corr = "fdr", sort.by = "fc",detect.lim = 1e-06,
    pr.cutoff = 1e-6, max.show = 50, plot.type = "quantile.box",
    panels = c("fc","auroc"),verbose = 1)

Arguments

siamcat	object of class siamcat-class
fn.plot	filename for the pdf-plot
color.scheme	valid R color scheme or vector of valid R colors (must be of the same length as the number of classes), defaults to 'RdYlBu'
alpha	float, significance level, defaults to 0.05
mult.corr	multiple hypothesis correction method, see p.adjust, defaults to "fdr"
sort.by	string, sort features by p-value ("p.val"), by fold change ("fc") or by prevalence shift ("pr.shift"), defaults to "fc"
detect.lim	float, pseudocount to be added before log-transformation of the data, defaults to 1e-06
pr.cutoff	float, cutoff for the prevalence computation, defaults to 1e-06
max.show	integer, how many associated features should be shown, defaults to 50
plot.type	string, specify how the abundance should be plotted, must be one of these: c("bean", "box", "quantile.box", "quantile.rect"), defaults to "quantile.box"
panels	vector, name of the panels to be plotted next to the log10- transformed abundances, possible entries are c("fc", "auroc", "prevalence"), defaults to c("fc", "auroc")
verbose	control output: 0 for no output at all, 1 for only information about progress and success, 2 for normal level of information and 3 for full debug information, defaults to 1

Value

Does not return anything, but produces an association plot

Examples

# Example data
    data(siamcat_example)
# since the whole pipeline has been run in the example data, exchange the
# normalized features with the original features
    siamcat_example <- reset.features(siamcat_example)

# Simple example
    check.associations(siamcat_example, './assoc_plot.pdf')

# Plot associations as bean plot
    check.associations(siamcat_example, './assoc_plot_bean.pdf',
    plot.type='bean')

# Plot assocations as box plot
# Additionally, sort by p-value instead of by fold change
    check.associations(siamcat_example, './assoc_plot_fc.pdf',
    plot.type='box', sort.by='p.val')

# Custom colors
    check.associations(siamcat_example, './assoc_plot_blue_yellow.pdf',
    plot.type='box', color.scheme=c('cornflowerblue', '#ffc125'))

KonradZych/SIAMCAT documentation built on May 17, 2019, 6:20 p.m.