From 7bbe4c444523f281d07f79aa8d0a4719668c3c80 Mon Sep 17 00:00:00 2001
From: Christoph Helma <helma@in-silico.ch>
Date: Sat, 20 Mar 2021 00:14:10 +0100
Subject: manuscript update

---
 scripts/pa-groups.R | 60 +++++++++++++++++++++++++++++++++--------------------
 1 file changed, 38 insertions(+), 22 deletions(-)

(limited to 'scripts/pa-groups.R')

diff --git a/scripts/pa-groups.R b/scripts/pa-groups.R
index 3c6ce2c..77b358b 100755
--- a/scripts/pa-groups.R
+++ b/scripts/pa-groups.R
@@ -1,33 +1,49 @@
 #!/usr/bin/env Rscript
 library(ggplot2)
 args = commandArgs(trailingOnly=TRUE)
-groups = read.csv(args[1],header=T)
-data = read.csv(args[2])
-for (i in c(2:10)) {
-  name = names(groups)[i]
-  cols = c(2:15)
-  group = data[groups[i] == 1,cols]
-  freq = 100*colSums(group,na.rm=TRUE)/colSums(!is.na(group))
-  algos = toupper(names(data)[cols])
-  algos = gsub("HIGH",'HC',algos)
-  algos = gsub(".CONFIDENCE",'',algos)
-  algos = gsub("\\.",'-',algos)
-  algos <- factor(algos,levels=rev(c(
-  "MP2D-LAZAR-ALL",
+freq = read.csv(args[1],header=T,quote="'",sep=",")
+
+models = factor(freq$Model,levels=rev(c(
   "MP2D-LAZAR-HC",
+  "MP2D-LAZAR-ALL",
   "MP2D-RF",
-  "MP2D-LR",
-  "MP2D-LR2",
+  "MP2D-LR-sgd",
+  "MP2D-LR-scikit",
   "MP2D-NN",
   "MP2D-SVM",
-  "CDK-LAZAR-ALL",
   "CDK-LAZAR-HC",
+  "CDK-LAZAR-ALL",
   "CDK-RF",
-  "CDK-LR",
-  "CDK-LR2",
+  "CDK-LR-sgd",
+  "CDK-LR-scikit",
   "CDK-NN",
   "CDK-SVM"
-  )))
-  plot = ggplot(data.frame(freq),aes(x=freq,y=algos)) + geom_bar(stat="identity") + xlab("% mutagenic") + ylab(element_blank()) + xlim(c(0,100))# + theme(axis.text.x = element_text(angle=90))
-  ggsave(paste("figures/",name,".png",sep=""))
-}
+)))
+
+colors <- c(
+"MP2D-LAZAR-HC" =  "#0072B2",
+"MP2D-LAZAR-ALL" = "#56B4E9",
+"MP2D-RF" =        "#009E73",
+"MP2D-LR-sgd" =        "#F0E442",
+"MP2D-LR-scikit" =       "#D55E00",
+"MP2D-NN" =        "#CC79A7",
+"MP2D-SVM" =       "#E69F00",
+"CDK-LAZAR-HC" =   "#0072B2",
+"CDK-LAZAR-ALL" =  "#56B4E9",
+"CDK-RF" =         "#009E73",
+"CDK-LR-sgd" =         "#F0E442",
+"CDK-LR-scikit" =        "#D55E00",
+"CDK-NN" =         "#CC79A7",
+"CDK-SVM" =        "#E69F00"
+)
+
+ggplot(freq,aes(Frequency,models,fill=models)) +
+  geom_bar(stat="identity",show.legend=F) +
+  xlab("% mutagenic") +
+  ylab(element_blank()) +
+  xlim(c(0,100)) +
+  scale_fill_manual(values = colors) +
+  facet_wrap(~PA.Group) +
+  theme_minimal() +
+
+ggsave(args[2])
-- 
cgit v1.2.3