plotROC {Anaquin}R Documentation

Create ROC plot

Description

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

Usage

plotROC(seqs, score, group, label, refGroup, title, legTitle)

Arguments

seqs

Sequin names

score

How to rank ROC points

group

How to group ROC points

label

True-positive (TP) or false positive (FP)

refGroup

Reference ratio groups

title

Label of the plot. Default to NULL.

legTitle

Title of the legend. Default to Ratio.

Details

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 function prints ROC plot and return it's AUC statistics.

Author(s)

Ted Wong t.wong@garvan.org.au

Examples

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
group <- abs(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

plotROC(seqs, score, group, label, title='ROC Plot', refGroup=0)

[Package Anaquin version 2.4.0 Index]