The basic idea of this plot is to show how counts data change while shifting a threshold-criterium.
At each given threshold the counts are plotted like a staggered bar-chart (or staggered histogram) but without vertical lines to illustrated the almost continuous change
from preceedig or following threshold-value.
Initially this plot was designed for showing the absolute count-data used when constructing roc-curves (eg using
the function summarizeForROC
of package wrProteo ).
The main input should furnish the panel of threshold as one column and the coresponding counts data as min 2 columns.
The threshold coumns gets specified using the argument threColumn
, the counts-data may either be specified using argument countsCol
or be searched using grep
using column-names containing the text given in argument varCountNa
with may be combined with
a fixed preceeding part given as argument fixedCountPat
.
staggerdCountsPlot(
roc,
threColumn = 1,
countsCol = NULL,
fixedCountPat = "n.pos.",
varCountNa = NULL,
sortAscending = TRUE,
vertLine = NULL,
col = NULL,
tit = NULL,
logScale = FALSE,
las.alph = 2,
displMaxSpec = TRUE,
silent = FALSE,
callFrom = NULL
)
(numeric matrix or data.frame) main input: one column with thresholds and multiple columns of assoicated count data
(integer or character) to specify the column with threshold-data, in typica proteomics benchmark studies this would be 'alph' (for the statistical test threshold)
(character of integer, min length=2) choice of column(s) with count-data in 'roc' to be used for display, if not NULL
will override alternative search of columns using 'varCountNa' and 'fixedCountPat'
(character) optional pattern to help identifying counts-data: if not NULL
it will be used as fixed part in column names to get pasted to varCountNa
.
In proteomics benchmark studies this would typically be 'n.pos.'
(character) alternative way to select the columns from 'roc': searched using grep
using column-names containing the text given in argument varCountNa
with may be combined with a fixed preceeding part given as argument fixedCountPat
In proteomics benchmark studies this would typically be the species-abbreciations (eg 'H','S','E')
(logical) decide if data should be sorted ascending or descending
(numeric) for optional vertical line, typically used to highlight alpha 0.05
(character) custom colors, see also par
(character) cutom title
(logical) display threshld values (x-axis) on log-scale
(numeric) orientation of label of alpha-cutoff, see also par
(logical) display on right side of figure max count value of contributing group species
(logical) suppress messages
(character) allow easier tracking of message(s) produced
plot only
Investigate count data prepared for plotting ROC curves : cumulative counts plot by species (along different statistical test thresholds). Note : Package wrProteo may be used to prepare input (matrix of ROC data).
ecdf
, for preparing input to ROC: function summarizeForROC
in package wrProteo
# NOT RUN {
set.seed(2019); test1 <- cbind(a=sample.int(n=7,size=50,repl=TRUE),
b=sample.int(n=11,size=50,repl=TRUE),c=sample.int(n=18,size=50,repl=TRUE))
test1 <- cbind(alph=seq(0,1,length.out=50),a=cumsum(test1[,1]),b=cumsum(test1[,2]),
c=cumsum(test1[,3]))
staggerdCountsPlot(test1,countsCol=c("a","b","c"))
## example below requires the package wrProteo
# }
Run the code above in your browser using DataCamp Workspace