tcga_boxplots: Partial Least Squares and predict second dataset
In graeberlab/small.cell.project: small.cell.project

Description Usage Arguments

Builds PLS model from training dataset and predicts second dataset Writes out predicted.scores of the second dataset Plots original PLS, projected samples only, and projected samples ontop of original PLS

1	tcga_boxplots(dir, pattern, output_string, output_folder = "./")

`output_folder`	the folder to output to, default is ./ i.e. current folder
`file`	file for X matirx
`sample.names`	Vector of sample names in X matrix
`sample.type`	vector of sample groups
`y.response`	numeric vector of response values in same order as samples
`comp`	number of components to compute
`scale`	default=T
`labels`	label the plot, default = T
`comp.x, comp.y`	comps to display
`title`	title of the plot
`file2`	file for test data matrix
`sample.names2`	Vector of sample names in 2nd dataset, if needed
`sample.type2`	Vector of sample types in 2nd dataset, if needed
`train_string`	string of training data to insert in file name of predicted scores
`test_string`	string of data being tested to insert in file name of predicted scores
`saveplot`	whether to save the plot, default is F
`savetype`	the type of plot to save,options are ".pdf" or ".png"
`w`	is width of plot to be saved
`h`	is height of plot to be saved
`legendname`	is the legend name
`plot_both`	if true plots both training and test set in color
`colpalette`	allows you to put in a color palette of form c("#F87660", "#39B600",....etc) to manually assign colors
`shape.palette`	allows you to put in a shape palette of form c(1, 3,....etc) to manually assign shapes
`varimax`	If T performs Varimax rotation,
`varimax.comp`	# of varimax components, kind of hacky, keep this # the same as # of comps. Will fix later.
`TCGA`	predicted files are from TCGA, barcodes separated by periods, so remove normal samples, default is FALSE