Upload new file

+setwd("/Users/larisams/Documents/LCG/Sem_4/Bioinformatics/CarlosMendez")
+library(ggplot2)
+# Automated analysis
+
+sent.dat<-as.data.frame(read.table("annotated-sentence-results.txt",header=TRUE))
+sent.dat$Proportion<-sent.dat$GeneDisease/sent.dat$Total
+sent.dat$constant<-apply(sent.dat,1,function(x){
+  if(x[3]==25){return("4")}
+  else if(x[3]==46){return("3")}
+  else if(x[3]==17){return("2")}
+  else{PConst<-return("0")}
+})
+
+g1<-ggplot(sent.dat,aes(x=Cluster,y=Proportion,fill=constant))+
+  geom_bar(aes(x=Cluster,y=Proportion),stat="identity")+
+  scale_fill_manual("Constant Partitions", values = c("4" = "#005b7a", "3" = "#DF0664", "2" = "#66CCCC" ,"0" = "#b2afaf"))+
+  facet_wrap(~Partition,ncol=3,nrow=2)+facet_grid(~Partition, labeller = label_context)+theme(legend.position="none")
+ggsave("Figure3.png",g1,width = 7.5, height = 5, dpi = 300, units = "in", device='png')
+
+g2<-ggplot(sent.dat,aes(x=Total,y=GeneDisease,fill=constant))+
+  geom_bar(aes(x=Total,y=GeneDisease),stat="identity")+
+  scale_fill_manual("Constant Partitions", values = c("4" = "#005b7a", "3" = "#DF0664", "2" = "#66CCCC" ,"0" = "#b2afaf"))+
+  facet_wrap(~Partition,ncol=3,nrow=2)+facet_grid(~Partition, labeller = label_context)+
+  xlab("Total sentences")+ylab("Gene disease sentences")+theme(legend.position="none")
+ggsave("Figure2.png",g2,width = 7.5, height = 5, dpi = 300, units = "in", device='png')
+
+# Manual analysis
+
+words.dat<-as.data.frame(read.table("word-counts.txt",header=TRUE))
+words.dat$Fibrosis<-words.dat$Fibrosis/words.dat$Total
+words.dat$TGF<-words.dat$TGF/words.dat$Total
+words.dat$constant<-sent.dat$constant[c(-1,-2)]
+
+g3<-ggplot(words.dat,aes(x=Cluster,y=Fibrosis,fill=constant))+
+  geom_bar(aes(x=Cluster,y=Fibrosis),stat="identity")+
+  scale_fill_manual("Constant Partitions", values = c("4" = "#005b7a", "3" = "#DF0664", "2" = "#66CCCC" ,"0" = "#b2afaf"))+
+  facet_wrap(~Partition,ncol=3,nrow=2)+facet_grid(~Partition, labeller = label_context)+theme(legend.position="none")
+ggsave("Figure7.png",g3,width = 7.5, height = 5, dpi = 300, units = "in", device='png')
+
+g4<-ggplot(words.dat,aes(x=Cluster,y=TGF,fill=constant))+
+  geom_bar(aes(x=Cluster,y=TGF),stat="identity")+
+  scale_fill_manual("Constant Partitions", values = c("4" = "#005b7a", "3" = "#DF0664", "2" = "#66CCCC" ,"0" = "#b2afaf"))+
+  facet_wrap(~Partition,ncol=3,nrow=2)+facet_grid(~Partition, labeller = label_context)+theme(legend.position="none")
+ggsave("Figure8.png",g4,width = 7.5, height = 5, dpi = 300, units = "in", device='png')
+
+
+
+
+
+
+
+
+
+
+