report update (incl metat polished assemblies) (issue #52)

workflow_report/rules/collect_data.smk

@@ -139,7 +139,7 @@ rule collect_quast:
             ifile_tool  = os.path.basename(os.path.dirname(ifile))
             ifile_df    = pandas.read_csv(ifile, sep="\t", header=0, index_col=0)
             ifile_df.rename(columns={"ASSEMBLY": ifile_tool}, inplace=True)
-            print(ifile_df)
+            # print(ifile_df)
             if summary is None:
                 summary = ifile_df.copy()
             else:

workflow_report/scripts/report.R

@@ -13,8 +13,10 @@ TABS$nanostats <- read_nanostats(snakemake@input$nanostats)
 
 TABS$mappability <- list()
 TABS$mappability$metag <- read_mappability(snakemake@input$mappability[1])
+TABS$mappability$metag$tool <- factor(TABS$mappability$metag$tool, ordered=TRUE, levels=ASM_TOOL_NAMES)
 if(length(snakemake@input$mappability) > 1){
     TABS$mappability$metat <- read_mappability(snakemake@input$mappability[2])
+    TABS$mappability$metat$tool <- factor(TABS$mappability$metat$tool, ordered=TRUE, levels=ASM_TOOL_NAMES)
 } else {
     TABS$mappability$metat <- NULL
 }
@@ -112,6 +114,7 @@ FIGS$quast <-
 
 TABS$plasflow <- read_plasflow(snakemake@input$plasflow)
 TABS$plasflow_melted <- reshape2::melt(TABS$plasflow, id.vars=c("label", "type", "tool"), variable.name="statstype")
+TABS$plasflow_melted$tool <- factor(TABS$plasflow_melted$tool, ordered=TRUE, levels=ASM_TOOL_NAMES)
 FIGS$plasflow <- list()
 
 for(stype in c("sum_pct")){
@@ -144,6 +147,8 @@ for(stype in c("sum_pct")){
 
 TABS$prodigal <- read_prodigal(snakemake@input$prodigal)
 TABS$prodigal_melted <- reshape2::melt(TABS$prodigal[,c("tool", "total", "partial")], id.vars=c("tool"))
+TABS$prodigal$tool <- factor(TABS$prodigal$tool, ordered=TRUE, levels=ASM_TOOL_NAMES)
+TABS$prodigal_melted$tool <- factor(TABS$prodigal_melted$tool, ordered=TRUE, levels=ASM_TOOL_NAMES)
 
 FIGS$prodigal <- list()
 FIGS$prodigal$counts <-
@@ -328,12 +333,14 @@ TABS$crispr_aggr_sa <- aggregate(
     by=list(crispr_tool=TABS$crispr$crispr_tool, asm_tool=TABS$crispr$asm_tool),
     FUN=sum
 )
+TABS$crispr_aggr_sa$asm_tool <- factor(TABS$crispr_aggr_sa$asm_tool, ordered=TRUE, levels=ASM_TOOL_NAMES)
 # aggregate: arrays per assembly
 TABS$crispr_aggr_aa <- aggregate(
     x=TABS$crispr$seq_id,
     by=list(crispr_tool=TABS$crispr$crispr_tool, asm_tool=TABS$crispr$asm_tool),
     FUN=length
 )
+TABS$crispr_aggr_aa$asm_tool <- factor(TABS$crispr_aggr_aa$asm_tool, ordered=TRUE, levels=ASM_TOOL_NAMES)
 
 FIGS$crispr <- list()
 crispr_theme <-

workflow_report/scripts/report.Rmd

@@ -26,11 +26,11 @@ knitr::kable(TABS$nanostats, caption='NanoStats')
 
 # Mappability
 
-```{r figures-mappability-metag, echo=FALSE, fig.width=10, fig.height=5, fig.cap='Mappability, metaG'}
+```{r figures-mappability-metag, echo=FALSE, fig.width=10, fig.height=7, fig.cap='Mappability, metaG'}
 print(FIGS$mappability$metag)
 ```
 
-```{r figures-mappability-metat, echo=FALSE, fig.width=10, fig.height=5, fig.cap='Mappability, metaT'}
+```{r figures-mappability-metat, echo=FALSE, fig.width=10, fig.height=7, fig.cap='Mappability, metaT'}
 if("metat" %in% names(FIGS$mappability)){
     print(FIGS$mappability$metat)
 }
@@ -46,7 +46,7 @@ grid.draw(FIGS$mash)
 # Assembly quality
 
 ```{r tables-quast, echo=FALSE}
-knitr::kable(TABS$quast, caption='QUAST')
+knitr::kable(t(TABS$quast), caption='QUAST')
 ```
 
 ```{r figures-quast, echo=FALSE, fig.width=10, fig.height=10, fig.cap='Assembly statistics w/ QUAST'}
@@ -55,7 +55,7 @@ print(FIGS$quast)
 
 # Plasmid prediction
 
-```{r figures-plasflow, echo=FALSE, fig.width=10, fig.height=5, fig.cap='Plasmid prediction w/ Plasflow'}
+```{r figures-plasflow, echo=FALSE, fig.width=10, fig.height=7, fig.cap='Plasmid prediction w/ Plasflow'}
 print(FIGS$plasflow$sum_pct)
 ```
 
@@ -108,6 +108,6 @@ FIGS$cdhit
 print(FIGS$rgi$total)
 ```
 
-```{r figures-rgi-overlap, echo=FALSE, fig.width=10, fig.height=5, fig.cap='AMR prediction w/ RGI'}
+```{r figures-rgi-overlap, echo=FALSE, fig.width=10, fig.height=7, fig.cap='AMR prediction w/ RGI'}
 plot_rgi_overlap(TABS$rgi, "strict", "ARO")
 ```
\ No newline at end of file

workflow_report/scripts/utils.R

@@ -170,19 +170,21 @@ plot_rgi_overlap <- function(df, ctype, col){
         # overlap order
         order.by="degree",
         decreasing=FALSE,
+        # number of sets to plot
+        nsets=length(ASM_TOOL_NAMES),
         # y-label title
         mainbar.y.label=sprintf("Intersection size (%s hits, %s)", ctype, col),
         # text size
-        text.scale = c(1.5, 1.5, 1.5, 1.5, 1.5, 1.5),
+        text.scale = c(1.2, 1.2, 1.2, 1.2, 1.2, 1.2)#,
         # 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, alpha=0.7))
-        )
+        # 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, alpha=0.7))
+        # )
     )
 }
 
@@ -210,14 +212,23 @@ plot_crispr_overlap <- function(df, asm_tool){
 
 # names
 ASM_TOOL_NAMES <- c(
+    # LR, SR, hybrid
     "flye"="Flye",
     "megahit"="MEGAHIT",
-    "metaspadeshybrid"="metaSPAdes (H)",
     "metaspades"="metaSPAdes",
-    "operams"="OPERA-MS"
+    "operams"="OPERA-MS",
+    "metaspadeshybrid"="metaSPAdes (H)",
+    # polishing w/ metaT (w/o LR)
+    "megahitmetatracon"="Racon(MEGAHIT + metaT)",
+    "metaspadesmetatracon"="Racon(metaSPAdes + metaT)",
+    "operamsmetatracon"="Racon(OPERA-MS + metaT)",
+    "metaspadeshybridmetatracon"="Racon(metaSPAdes (H) + metaT)"
 )
 # colors
-ASM_TOOL_COLORS <- ggsci::pal_nejm("default", alpha=1)(length(ASM_TOOL_NAMES))
+ASM_TOOL_COLORS <- c(
+    ggsci::pal_nejm("default", alpha=1)(5),
+    ggsci::pal_nejm("default", alpha=0.6)(5)[2:5]
+)
 names(ASM_TOOL_COLORS) <- ASM_TOOL_NAMES
 
 ###############