figures: added mmseqs2 fig

workflow_figures/Snakefile

@@ -30,6 +30,7 @@ rule all:
         fig_quast="figures/fig_quast.pdf",
         fig_mash="figures/fig_mash.pdf",
         fig_prodigal="figures/fig_prodigal.pdf",
+        fig_mmseqs2="figures/fig_mmseqs2.pdf",
         fig_gdb_rgi_aro3004454_flye="figures/fig_gdb_rgi_aro3004454_flye.pdf"
 
 rule fig_mappability:
@@ -106,6 +107,24 @@ rule fig_prodigal:
     script:
         os.path.join(SRC_DIR, "fig_prodigal.R")
 
+rule fig_mmseqs2:
+    input:
+        zymo=os.path.join(config["samples"]["zymo"], "analysis/mmseqs2/summary.tsv"),
+        nwc=os.path.join(config["samples"]["nwc"], "analysis/mmseqs2/summary.tsv"),
+        gdb=os.path.join(config["samples"]["gdb"], "analysis/mmseqs2/summary.tsv"),
+        rumen=os.path.join(config["samples"]["rumen"], "analysis/mmseqs2/summary.tsv")
+    output:
+        pdf="figures/fig_mmseqs2.pdf"
+    log:
+        "figures/fig_mmseqs2.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_mmseqs2.R")
+
 rule fig_gdb_rgi_aro3004454_flye_metaT:
     input:
         os.path.join(config["samples"]["gdb"], "mapping/metat/lr/flye/ASSEMBLY.POLISHED.sr.cov.perbase")

workflow_figures/scripts/fig_mmseqs2.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$samples)){
+    testit::assert(sname %in% names(snakemake@input))
+    TABS[[sname]] <- read_mmseqs(snakemake@input[[sname]])
+}
+
+############################## PLOT
+
+# individual plots
+FIGS <- list()
+for(sname in names(snakemake@config$samples)){
+    FIGS[[sname]] <-
+        as.ggplot(plot_mmseqs2_overlap(TABS[[sname]])) +
+    labs(
+        title=LETTERS[which(names(TABS) == sname)],
+        subtitle=SAMPLE_NAMES[sname]
+    )
+}
+
+# combine plots
+FIGS_ALL <- Reduce("+", FIGS) + plot_layout(ncol=2) #+ plot_annotation(tag_levels='A')
+
+# PDF
+pdf(snakemake@output$pdf, width=20, height=14)
+print(FIGS_ALL)
+dev.off()
\ No newline at end of file