prelim set-up: snakemake file and config, src for mmseq fig (issue #16)

figures/figures.smk

+##################################################
+# Config
+##################################################
+configfile: "figures.yml"
+
+##################################################
+# Python modules
+##################################################
+import os
+
+##################################################
+# TARGETS
+##################################################
+FIG_MMSEQ_UPSETR = os.path.join(config["output"], config["fig_mmseq_upsetr"]["output"])
+
+##################################################
+# RULES
+##################################################
+rule all:
+    input:
+        FIG_MMSEQ_UPSETR
+
+rule mmseq_upsetr:
+    input:
+        overlap_sizes=config["fig_mmseq_upsetr"]["overlap_sizes"]
+    output:
+        pdf=FIG_MMSEQ_UPSETR
+    log:
+        FIG_MMSEQ_UPSETR + ".log"
+    params:
+        utils=config["utils"],
+        width=config["fig_mmseq_upsetr"]["width"],
+        height=config["fig_mmseq_upsetr"]["height"]
+    conda:
+        "envs/r.yml"
+    script:
+        config["fig_mmseq_upsetr"]["script"]
\ No newline at end of file

figures/figures.yml

+# general
+output: "output" # output directory
+utils: "src/utils.R" # path to utils.R
+
+# figures
+fig_mmseq_upsetr:
+    script: "src/mmseq_upsetr.R"
+    overlap_sizes: "data/overlap_sizes.txt"
+    output: "fig_mmseq_upsetr.pdf"
+    width: 7
+    height: 5
\ No newline at end of file

figures/src/mmseq_upsetr.R

+#!/usr/bin/Rscript
+
+## LOG FILE
+sink(file=file(snakemake@log[[1]], open="wt"), type="message")
+
+## NOTE
+# UpSetR
+# https://cran.r-project.org/web/packages/UpSetR/vignettes/basic.usage.html
+
+## IMPORT
+suppressMessages(library(testit))
+suppressMessages(library(UpSetR))
+# custom
+source(snakemake@params$utils)
+
+## DATA
+# overlap/intersection sizes
+overlap_sizes <- read.csv(
+    file=snakemake@input$overlap_sizes,
+    sep=" ",
+    header=FALSE,
+    stringsAsFactors=FALSE
+)
+# check: all expected tools are there
+testit::assert(all(sapply(names(TOOL_NAMES), function(x){ x %in% overlap_sizes$V2 })))
+# process names
+for(tname in names(TOOL_NAMES)){
+    overlap_sizes$V2 <- sub(tname, TOOL_NAMES[tname], overlap_sizes$V2)
+}
+overlap_sizes$V2 <- gsub("_", "&", overlap_sizes$V2)
+# create input for UpSetR
+overlap_names <- overlap_sizes$V2
+overlap_sizes <- overlap_sizes$V1
+names(overlap_sizes) <- overlap_names
+
+## PLOT
+pdf(snakemake@output$pdf, width=snakemake@params$width, height=snakemake@params$height)
+UpSetR::upset(
+    data=UpSetR::fromExpression(overlap_sizes),
+    # overlap order
+    order.by="degree",
+    decreasing=FALSE,
+    # colors
+    # queries=lapply(
+    #     TOOL_NAMES,
+    #     function(x){ list(query=UpSetR::intersects, params=list(x), color=TOOL_COLORS$proc[[x]], active=TRUE) }
+    # ),
+    set.metadata=list(
+        data=data.frame(
+            sets=TOOL_NAMES,
+            Tool=TOOL_NAMES,
+            stringsAsFactors=FALSE
+        ), 
+        plots=list(list(type="matrix_rows", column="Tool", colors=TOOL_COLORS$proc, alpha=0.7))
+    )
+)
+dev.off()

figures/src/utils.R

+#!/usr/bin/Rscript
+
+suppressMessages(library(ggsci)) # colors
+
+# how the tool names should be changed
+TOOL_NAMES <- c(
+    "flye"="Flye",
+    "megahit"="MEGAHIT",
+    "metaspades_hybrid"="metaSPAdes (hybrid)",
+    "metaspades"="metaSPAdes"
+)
+# tool colors
+TOOL_COLORS <- list()
+TOOL_COLORS$raw  <- ggsci::pal_nejm("default", alpha=1)(4); names(TOOL_COLORS$raw) <- names(TOOL_NAMES)
+TOOL_COLORS$proc <- TOOL_COLORS$raw; names(TOOL_COLORS$proc) <- TOOL_NAMES