run_comparison: Run analysis in streamlined approach

In kazeera/hourglass: This Package Provides Visualizations of Annotated Numeric Data

Run analysis in streamlined approach

Description

Create the plots for each comparison. Run the same analysis on dataset object (ds parameter) and transformed values (ds.imp parameter, optional).

Usage

run_comparison(
  ds,
  rowAnns,
  colAnns = NA,
  output_folder = ".",
  ds.imp = NULL,
  feat_sets = NULL,
  var_colors = NULL,
  gradient_palette = "RdBu",
  corr_method = "pearson",
  pval.test = "t.test",
  pval.label = "p.signif",
  boxplot_log10_y = F,
  FC.method = "divide",
  paired_analysis_column = NA,
  do_paired_analysis = F,
  make.QC.param = F,
  make.QC.feature = F,
  make.feat.plots = F,
  discrete_stacked_params = NULL,
  make.het.plot = F,
  make.indiv.boxplot = F,
  make.overview.boxplot = F,
  make.heatmap = F,
  make.corrplot = F,
  make.overview.corrscatt = F,
  make.indiv.corrscatt = F,
  make.barplot = F,
  make.FC.pval.plot = F,
  save_table = F
)

Arguments

ds	A dataset object (a list with vals, rowAnn, colAnn, comparison, name).
rowAnns	A character vector of 1-2 column names in ds$rowAnn. c(MainComparison, Subgroup)
colAnns	A character vector of 1-2 column names in ds$colAnn. c(Parameter, Feature/Stain/Gene)
output_folder	The main output folder for all custom analysis plots and boxplots for by.parameter and by.feature analysis
ds.imp	A dataset object similar to ds with imputed or another transformation values.
feat_sets	A list of 2 data frames for feature sets and parameters.
var_colors	A named vector with colors as values and annotations/groups as names.
gradient_palette	RColorBrewer palette name for gradients (e.g. heatmap, correlation plots). See RColorBrewer::display.brewer.all() for all options.
corr_method	Method for correlation (one of "pearson","spearman","kendall").
pval.test	Which two-samples testing should be used? String corresponding to "method" parameter in stat_compare_means. Allowed values are "t.test" and "wilcox.test".
pval.label	How to display p-values? String corresponding to "label" parameter in stat_compare_means. Allowed values are "p.signif" (stars) and "p.format" (number).
boxplot_log10_y	Log10 the values on y axis for boxplots and patient paired slopegraphs? Logical (T/F). Default is FALSE.
FC.method	Fold change computation method to use, either "divide" (for non-transformed values) or "subtract" (for log2-transformed values)
paired_analysis_column	column name in ds$rowAnn to create paired analysis plots for, e.g. PatientID if ds is data for all cores
do_paired_analysis	makes plots to look at subgroup differences within the same patient. Will only run if paired_analysis_column is specified.
make.QC.param, make.QC.feature, make.feat.plots, make.het.plot, make.indiv.boxplot, make.overview.boxplot, make.heatmap, make.corrplot, make.overview.corrscatt, make.indiv.corrscatt, make.barplot, make.FC.pval.plot	Logicals (TRUE/FALSE) indicating whether to make these plots. Note: make.indiv.corrscatt = T takes a long time.
discrete_stacked_params	parameter names to search for in colAnn1 to make discrete stacked barplots, e.g. "Het.Score"
save_table	Print MainComparison + ID data to csv file.
run	A one row data frame or list object with logicals for what to run, names: make.boxplots, make.paired.boxplots, make.heatmaps, make.surv.curve

kazeera/hourglass documentation built on April 5, 2025, 7:18 a.m.