Figures valentina

figures/src/fig_rgi.R

@@ -11,6 +11,7 @@ suppressMessages(library(testit))
 suppressMessages(library(UpSetR))
 suppressMessages(library(ggplot2))
 suppressMessages(library(reshape2))
+suppressMessages(library(UpSetR))
 # custom
 source(snakemake@params$utils)
 
@@ -22,12 +23,14 @@ stats <- read.csv(
     stringsAsFactors=FALSE,
     check.names=FALSE
 )
+# change names
+colnames(stats) <- sapply(colnames(stats), function(x){ ifelse(x %in% names(ASM_TOOL_NAMES), ASM_TOOL_NAMES[x], x) })
 # reshape
 stats_melted <- reshape2::melt(stats, id.vars=c("label", "col", "type"), variable.name="tool", value.name="count")
 # change names
-stats_melted$tool <- ASM_TOOL_NAMES[as.character(stats_melted$tool)]
+# stats_melted$tool <- ASM_TOOL_NAMES[as.character(stats_melted$tool)]
 
-## PLOT
+## PLOT: counts
 my_theme <- 
     theme_bw() +
     theme(
@@ -60,20 +63,6 @@ for(ctype in unique(stats_melted$type)){
             ) +
             my_theme
     }
-    # Total counts
-    # df <- stats_melted[stats_melted$type == ctype,]
-    # df <- aggregate(df$count, by=list(tool=df$tool), FUN=sum)
-    # plots[[sprintf("%s_%s", ctype, "total")]] <-
-    #     ggplot(data=df, aes(x=tool, y=x, fill=tool)) +
-    #     geom_col() +
-    #     scale_fill_manual(values=ASM_TOOL_COLORS$notmeth, guide=NULL) +
-    #     labs(
-    #         # title="",
-    #         subtitle=sprintf("Hits: %s", ctype),
-    #         x="",
-    #         y="Total hits"
-    #     ) +
-    #     my_theme
 }
 # Total counts
 df <- stats_melted[stats_melted$col == "ARO",]
@@ -89,7 +78,38 @@ plots[[sprintf("%s_%s", ctype, "total")]] <-
     ) +
     my_theme
 
-## PDF
+## PLOT: intersection &  PDF
+plot_overlap <- function(df, ctype, col){
+    df <- df[df$type == ctype & df$col == col,]
+    df_list <- lapply(ASM_TOOL_NAMES_NOTMETH, function(x){ df[df[,x] > 0,"label"] })
+    names(df_list) <- ASM_TOOL_NAMES_NOTMETH[names(df_list)]
+    UpSetR::upset(
+        data=UpSetR::fromList(df_list),
+        # overlap order
+        order.by="degree",
+        decreasing=FALSE,
+        # y-label title
+        mainbar.y.label=sprintf("Intersection size (%s hits, %s)", ctype, RGI_NAMES$col[col]),
+        # text size
+        text.scale = c(2, 2, 1.5, 1.5, 2, 2),
+        # colors
+        set.metadata=list(
+            data=data.frame(
+                sets=names(df_list),
+                Tool=names(df_list),
+                stringsAsFactors=FALSE
+            ), 
+            plots=list(list(type="matrix_rows", column="Tool", colors=ASM_TOOL_COLORS$notmeth, alpha=0.7))
+        )
+    )
+}
+
+
 pdf(snakemake@output$pdf, width=snakemake@params$width, height=snakemake@params$height)
+# Count plots
 for(pp in plots){ print(pp) }
+# Intersection plot
+plot_overlap(stats, "all", "ARO")
+plot_overlap(stats, "nudged", "ARO")
+plot_overlap(stats, "strict", "ARO")
 dev.off()