Create ROC plot

Share:

Description

Create a receiver operating characteristic (ROC) plot at various threshold settings.

Usage

1
plotROC(data, refRats, title, legTitle, ...)

Arguments

data

Anaquin dataset created by AnaquinData. It needs to define information in Details.

refRats

Reference ratio groups

title

Label of the plot. Default to NULL.

legTitle

Title of the legend. Default to Ratio.

...

Reserved for internal testing

Details

plotROC requires the following data inputs from AnaquinData.

seqs List of sequin identifiers (eg. R2_11_2)
label Classified labels ('TP' or 'FP')
score How the ROC points should be ranked
ratio Expected ratio; eg: expected log-fold ratio

Create a receiver operating characteristic (ROC) plot at various threshold settings. The true positive rate (TPR) is plotted on the x-axis and false positive rate (FPR) is plotted on the y-axis.

The function requires a scoring threshold function, and illustrates the performance of the data as the threshold is varied. Common scoring threshold include p-value, sequencing depth and allele frequency, etc.

ROC plot is a useful diagnostic performance tool; it provides tools to select possibly optimal models and to discard suboptimal ones. In particularly, the AUC statistics indicate the performance of the model relatively to a random experiment (AUC 0.5).

Value

The functions does not return anything but it prints a ROC plot and it's AUC statistics.

Author(s)

Ted Wong t.wong@garvan.org.au

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
library(Anaquin)

#
# Data set generated by DESeq2 and Anaquin. described in Section 5.6.3.3 of
# the user guide.
#
data(UserGuideData_5.6.3)

# Sequin names
seqs <- row.names(UserGuideData_5.6.3)

# Expected log-fold
ratio <- UserGuideData_5.6.3$ExpLFC

# How the ROC curves are ranked
score <- 1-UserGuideData_5.6.3$Pval

# Classified labels (TP/FP)
label <- UserGuideData_5.6.3$Label

anaquin <- AnaquinData(analysis='PlotROC',
                           seqs=seqs,
                          ratio=ratio,
                          score=score,
                          label=label)

plotROC(anaquin, title='ROC Plot', refRats=0)