plotROC: Create ROC plot

In Anaquin: Statistical analysis of sequins

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 [e-mail]

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)

Example output

Loading required package: ggplot2


|   |    AUC|
|--:|------:|
|  4| 0.9062|
|  3| 0.8939|
|  2| 0.7955|
|  1| 0.6713|
$AUC
       AUC
4 4 0.9062
3 3 0.8939
2 2 0.7955
1 1 0.6713

Anaquin documentation built on Nov. 8, 2020, 5:23 p.m.