plotContCat: plot Cont-Cat variable
In nickytong/GenAnalysis: Genomic Analysis

for cont~cat, we need a boxplot with jittered points. An ANOVA p value is needed. A summary of anova model can be available.

plotContCat(x, y, main, tag, col = NULL, colI = NULL, colVar = NULL,
  colVar_pal = NULL, type = "jitter", xlab = "", ylab = "",
  useRankTest = FALSE, print = TRUE, ignoreNA = TRUE, order = FALSE,
  decreasing_order = FALSE, theme_bw = TRUE, ylim = NULL,
  jitter_w = 0.2, nybins = 100, barwidth = 0.2, barcol = NULL,
  cex.axis = NULL, lx = NULL, ly = NULL, decreasing = TRUE,
  labels = NULL, near0 = 0.01, showLegend = TRUE, width = 0.8,
  size = 4, sizeForce = NULL, sizex = 12, wrap_width_x = Inf,
  sizey = 12, sizetitlerel = 0.8, sizetitleXrel = 1, sizetitleYrel = 1,
  anglex = NULL, angley = 0, shape = 20, plot = TRUE, seed = 1000,
  expand = TRUE, base = NULL, FCbyRatio = FALSE, levels = NULL,
  saveTable = FALSE, pathTable0 = NULL, pathTable = NULL)

`x`	required to be a factor due to TukeyHSD(aov(y~x, data=mydf)) requires factor x. It is also necessary for the geom_boxplot otherwise a single box would be plotted when x is treated as continuous variable
`y`	continuous vector
`tag`	used to add a tag to the title
`col`	a vector of color specifying the color of box and points. Color correspondance is through scale_color_manual(breaks=levels(x), values=col)
`colI`	a vector of color of length(x) specifying the color of points.
`type`	plot type, either 'jitter' by geom_jitter (y is fixed) , 'dotplot' by geom_dotplot (y is binned), or 'barplot' by geom_barplot()
`xlab`	xlab
`ylab`	ylab
`useRankTest`	whether to use wilcox rank test in place of t test; only relevant if only two categories is found.
`print`	print text on console for test result
`ignoreNA`	whether to exclude the bar of NAs in the data. Default is to leave it there. Notice this will not affect the P value since NAs is not treated as a separate group but deleted
`order`	whether order factor by median (so that to match with the median bar in boxplot) of y values
`theme_bw`	logical if bw theme is imposed
`nybins`	number of bins on y axis if type=='dotplot'; this controls how y is grouped; dot size is determined through size;
`barwidth`	barwidth when type='barplot'
`barcol`	barcol when type='barplot'
`cex.axis`	size of index text when type='barplot'
`lx,`	ly legend position when type='barplot'
`decreasing`	how the factor level are ordered, decreasing TRUE or FALSE
`labels`	index text, e.g. cell line labels to show. default is integer index.
`near0`	when type='barplot', bar at 0 will not be shown due to 0 length; we can use a bar hight of near0 to represent it.
`width`	boxplot width; default is 1, which should be the ggplot2 default
`size`	points size
`sizex`	xaxis text size
`wrap_width_x`	use str_wrap to break long axis into multiple lines; default is not to break by setting this as Inf.
`sizey`	yaxis text size
`anglex`	xaxis text rotation
`angley`	yaxis text rotation
`shape`	points shape
`plot`	if FALSE, no figure will be generated. If TRUE, figure and print will be added
`seed`	jitter is random; add seed to make it reproducible
`expand`	whether to expand the size and axis title (useful if figure is to be copied in slides with multiple panels)
`base`	only relevant if categorical variable is binary which is used to calculate fold change. Default is Null, which will not enable FC calculation.
`FCbyRatio`	only active if two groups present, if specified as TRUE, need also to specify base=1 to be active
`levels`	only relevant if categorical variable is binary which is used to calculate fold change and mean diff. Default is NULL, which will be using alphabetical order, handled by ttestP()
`saveTable`	whether to save anova result in a csv file
`pathTable`	path to the saved csv file. default is NULL, which will be set as file.path(getwd(), 'OutputData/scatterTables')
`decreasing`	order direction of the bars

modified on 2014/01/29: x can be cont, y can be cat. This can be implemented by cord_flip() modified on 2014/02/13: add the option ignoreNA. This happens when we want to remove the bar of NA values. Also, n_unique(x) should not count NA as a level when claiming a test name. the reason is that when only 2 unique levels plus NA, we should say this is a t.test

nickytong/GenAnalysis documentation built on July 20, 2019, 8:57 a.m.