figures: added rgi fig

workflow_figures/Snakefile

@@ -33,6 +33,7 @@ rule all:
         fig_diamond_db="figures/fig_diamond_db.pdf",
         fig_mmseqs2="figures/fig_mmseqs2.pdf",
         fig_barrnap="figures/fig_barrnap.pdf",
+        fig_rgi=["figures/fig_rgi_counts.pdf", "figures/fig_rgi_overlap.pdf"],
         fig_gdb_rgi_aro3004454_flye="figures/fig_gdb_rgi_aro3004454_flye.pdf"
 
 rule fig_mappability:
@@ -145,6 +146,25 @@ rule fig_barrnap:
     script:
         os.path.join(SRC_DIR, "fig_barrnap.R")
 
+rule fig_rgi:
+    input:
+        zymo=os.path.join(config["samples"]["zymo"], "annotation/rgi/summary.tsv"),
+        nwc=os.path.join(config["samples"]["nwc"], "annotation/rgi/summary.tsv"),
+        gdb=os.path.join(config["samples"]["gdb"], "annotation/rgi/summary.tsv"),
+        rumen=os.path.join(config["samples"]["rumen"], "annotation/rgi/summary.tsv")
+    output:
+        pdf_c="figures/fig_rgi_counts.pdf",
+        pdf_o="figures/fig_rgi_overlap.pdf"
+    log:
+        "figures/fig_rgi.log"
+    params:
+        const=os.path.join(SRC_DIR, "const.R"),
+        utils=os.path.join(SRC_DIR, "utils.R")
+    conda:
+        os.path.join(ENV_DIR, "r.yaml")
+    script:
+        os.path.join(SRC_DIR, "fig_rgi.R")
+
 rule fig_mmseqs2:
     input:
         zymo=os.path.join(config["samples"]["zymo"], "analysis/mmseqs2/summary.tsv"),

workflow_figures/scripts/fig_rgi.R

+#!/usr/bin/Rscript
+
+# Figure: QUAST statistics
+
+############################## LOG
+sink(file=file(snakemake@log[[1]], open="wt"), type=c("output", "message"))
+
+############################## LIBS
+suppressMessages(library(testit)) # assertions
+suppressMessages(library(dplyr)) # tables
+suppressMessages(library(patchwork)) # combine plots
+suppressMessages(library(ggplotify)) # for as.ggplot
+
+# custom
+source(snakemake@params$const)
+source(snakemake@params$utils)
+
+############################## CONFIG
+
+# assemblers
+ASM_NAMES_CONFIG <- c(
+    snakemake@config$assemblers$lr,
+    snakemake@config$assemblers$sr,
+    snakemake@config$assemblers$hy
+)
+ASM_TOOL_NAMES  <- ASM_TOOL_NAMES[ASM_NAMES_CONFIG]
+ASM_TOOL_COLORS <- ASM_TOOL_COLORS[ASM_NAMES_CONFIG]; names(ASM_TOOL_COLORS) <- ASM_TOOL_NAMES[names(ASM_TOOL_COLORS)]
+ASM_TOOL_SHAPES <- ASM_TOOL_SHAPES[ASM_NAMES_CONFIG]; names(ASM_TOOL_SHAPES) <- ASM_TOOL_NAMES[names(ASM_TOOL_SHAPES)]
+
+############################## DATA
+
+# individual tables
+TABS <- list()
+for(sname in names(snakemake@config$samples)){
+    testit::assert(sname %in% names(snakemake@input))
+    TABS[[sname]] <- read_rgi(snakemake@input[[sname]]) 
+}
+
+############################## PLOT
+
+# individual plots
+FIGS_c <- list()
+FIGS_o <- list()
+for(sname in names(snakemake@config$samples)){
+    if(nrow(TABS[[sname]]) > 0){
+        FIGS_c[[sname]] <-
+            plot_rgi_counts_total(proc_rgi(TABS[[sname]])$aggr) +
+            labs(
+                title=LETTERS[which(names(TABS) == sname)],
+                subtitle=SAMPLE_NAMES[sname]
+            )
+        FIGS_o[[sname]] <-
+            as.ggplot(plot_rgi_overlap(TABS[[sname]], "strict", "ARO")) +
+            labs(
+                title=LETTERS[which(names(TABS) == sname)],
+                subtitle=SAMPLE_NAMES[sname]
+            )
+    }
+}
+
+# combine plots
+FIGS_ALL_c <- Reduce("+", FIGS_c) + plot_layout(ncol=2)
+FIGS_ALL_o <- Reduce("+", FIGS_o) + plot_layout(ncol=2)
+
+# PDF
+pdf(snakemake@output$pdf_c, width=20, height=14)
+print(FIGS_ALL_c)
+dev.off()
+
+pdf(snakemake@output$pdf_o, width=20, height=14)
+print(FIGS_ALL_o)
+dev.off()
\ No newline at end of file