Merge pull request #35 from mariecrane/stacked_bar

Add stacked bar plot to viz module

dev_environment.yml

@@ -16,5 +16,7 @@ dependencies:
 - r-ggplot2==3.0.*
 - r-base==3.5.1
 - r-scico=1.1.0
+- r-dplyr=0.8.0.1
+- r-tidyr=0.8.3
 - r-jsonlite=1.5.*
 - libiconv 1.15.*

metaquantome/cli.py

@@ -216,7 +216,7 @@ def parse_args_cli():
                              help='Perform paired tests.')
 
     # ---- METAQUANTOME VIZ ---- #
-    parser_viz.add_argument('--plottype', '-p', required=True, choices=['bar', 'volcano', 'heatmap', 'pca', 'ft_dist'],
+    parser_viz.add_argument('--plottype', '-p', required=True, choices=['bar', 'volcano', 'heatmap', 'pca', 'ft_dist', 'stacked_bar'],
                             help="Select the type of plot to generate.")
     parser_viz.add_argument('--img', required=True,
                             help='Path to the PNG image file (must end in ".png").')
@@ -231,19 +231,19 @@ def parse_args_cli():
     parser_viz.add_argument('--tabfile', default=None,
                             help="Optional. File to write plot table to.")
 
-    bar = parser_viz.add_argument_group('Arguments for barplots - both total taxonomy peptide intensity ("bar") and ' +
-                                        'function-taxonomy interaction distributions ("ft_dist")')
+    bar = parser_viz.add_argument_group('Arguments for barplots - including total taxonomy peptide intensity ("bar"), function-taxonomy ' +
+                                        'interaction distributions ("ft_dist"), and stacked taxonomy bar plots ("stacked_bar")')
     bar.add_argument('--meancol',
                      help="(Tax bar and FT dist). Mean intensity column name for desired experimental conditio.")
     bar.add_argument('--nterms', default='5',
-                     help="(Tax bar and FT dist). Number of taxa or functional terms to display. The default is 5.")
+                     help="(Tax bar, FT dist, and stacked bar). Number of taxa or functional terms to display. The default is 5.")
     bar.add_argument('--barcol', type=check_col_range, default="6",
                      help="(Tax bar and FT dist). Color for the bar fill. The color vector in R is " +
                           'c("dodgerblue", "darkorange", "yellow2", "red2", "darkviolet", "black"), ' +
                           ' so providing a 1 will give the "dodgerblue" color. These same colors are also used in the ' +
                           ' heatmap and PCA plot, so the colors can be tweaked to match. ')
     bar.add_argument('--target_rank',
-                     help="(Tax bar and FT dist). Taxonomic rank to restrict to in the plot. ")
+                     help="(Tax bar, FT dist, and stacked bar). Taxonomic rank to restrict to in the plot. ")
     bar.add_argument('--target_onto', choices=["mf", "bp", "cc"],
                      help="(Function and FT dist bar only) " +
                           "Ontology to restrict to, for function distribution.")

metaquantome/modules/run_viz.py

@@ -7,7 +7,7 @@
 
 
 def run_viz(plottype, img, infile, strip=None,
-            mode=None, meancol=None, nterms='5', target_rank=None, barcol=6,  # barplot
+            mode=None, meancol=None, nterms='5', target_rank=None, barcol=6,  # barplot, stacked_bar
             textannot=None, fc_name=None, flip_fc=False, gosplit=False,  # volcano
             sinfo=None, filter_to_sig=False, alpha='0.05',  # heatmap
             calculate_sep=False,  # pca
@@ -35,11 +35,16 @@ def run_viz(plottype, img, infile, strip=None,
         all_intcols_str = ','.join(samp_grps.all_intcols)
         json_dump = json.dumps(samp_grps.sample_names)
         cmd += [all_intcols_str, json_dump, calculate_sep, width, height, strip]
-    if plottype == "ft_dist":
+    elif plottype == "ft_dist":
         cmd += [whichway, name, id, meancol, nterms, width, height,
                 target_rank, target_onto, barcol, tabfile]
+    if plottype == "stacked_bar":
+        samp_grps = SampleGroups(sinfo)
+        all_intcols_str = ','.join(samp_grps.all_intcols)
+        json_dump = json.dumps(samp_grps.sample_names)
+        cmd += [all_intcols_str, json_dump, nterms, target_rank, width, height, tabfile]
     else:
-        ValueError("Wrong plot type. Must be bar, volcano, heatmap, ft_dist, or pca.")
+        ValueError("Wrong plot type. Must be bar, volcano, heatmap, ft_dist, stacked_bar, or pca.")
     # ensure that all elements are strings (even booleans, etc)
     cmd_string = [str(elem) for elem in cmd]
 

metaquantome/modules/viz.R

@@ -4,8 +4,11 @@ options(stringsAsFactors = FALSE, message=FALSE, warnings=FALSE)
 ####### ==================== #######
 #              LIBRARIES           #
 ####### ==================== #######
+suppressMessages(library(dplyr))
+suppressMessages(library(tidyr))
 suppressWarnings(suppressMessages(library(ggplot2)))
 suppressMessages(library(gplots))
+suppressMessages(library(RColorBrewer))
 suppressMessages(library(jsonlite))
 suppressMessages(library(stringr))
 
@@ -626,6 +629,84 @@ ft_dist_cli <- function(args){
     				  int_barcol=barcol, tabfile=tabfile)
 }
 
+####### ==================== #######
+#           STACKED BAR            #
+####### ==================== #######
+
+mq_stacked <- function(img, df, all_intcols, json_dump, nterms, target_rank, width, height, tabfile){
+  # df is the dataframe after stat
+  # nterms is the number of taxa to show
+
+  grp_list <- fromJSON(json_dump)
+
+  grp_df <- grp_list %>% 
+    as.data.frame() %>% 
+    pivot_longer(cols = 1:ncol(.), names_to = "samplegroup", values_to = "sample")
+
+  # parse out sample groups, exponentiate, calculate relative abundance
+  dat <- df %>% 
+    pivot_longer(all_intcols, names_to = "sample", values_to = "abundance") %>% 
+    full_join(grp_df) %>% 
+    mutate(replicate = str_replace(string = sample, pattern = samplegroup, replacement = "")) %>% 
+    replace_na(list(abundance = 0)) %>%
+    mutate(abundance = 2^abundance) %>%
+    select(sample, samplegroup, replicate, id, name, rank, abundance) %>% 
+    filter(rank == target_rank) %>% 
+    group_by(sample) %>% 
+    mutate(abundance = 100*abundance/sum(abundance))
+
+  # reorder taxa levels for plotting
+  taxa_levels <- names(sort(tapply(dat$abundance, dat$name, sum)))
+
+  # collapse less abundang terms into "Other" if terms exceed desired terms
+  if(length(unique(dat$name))>nterms){
+    topn <- tail(taxa_levels, nterms)
+    dat<-dat %>%
+      mutate(name = factor(ifelse(name %in% topn, name, "Other"))) %>% 
+      group_by(sample, samplegroup, replicate, name) %>% 
+      summarise(abundance = sum(abundance)) %>% 
+      group_by(sample)
+    taxa_levels <- names(sort(tapply(dat$abundance, dat$name, sum)))
+    # reorder taxa levels for plotting, to have "Other" on top
+    other_index <- as.numeric(which(taxa_levels == "Other"))
+    taxa_levels <- c("Other", taxa_levels[1:(other_index-1)], taxa_levels[(other_index+1):length(taxa_levels)])
+  }
+
+  # make stacked bar plot
+  fig <- dat %>%     
+    ggplot(aes(x=replicate, y=abundance, fill=factor(name, levels = taxa_levels)))+
+    geom_bar(position="stack", stat = "identity") +
+    facet_grid(cols = vars(samplegroup)) +
+    labs(x= "Sample", y="Relative Abundance")+
+    scale_fill_brewer(name="Taxa", palette = "Set1")
+
+  # write tabular file
+  write.table(x = dat, file = tabfile, quote = FALSE, row.names = FALSE)
+
+  # save plot
+  ggsave(file=img, width=width, height=height, units="in", dpi=300)
+
+}
+
+stacked_cli <- function(args){
+
+  img <- args[2]
+  infile <- args[3]
+  df <- read_result(infile)
+
+  # split all_intcols from SampleGroups(), for samp_columns vector
+  all_intcols <- unlist(strsplit(get_all_intcols(args[4]), split=","))
+
+  # other args
+  json_dump <- args[5]
+  nterms <- as.numeric(args[6])
+  target_rank <- toString(args[7])
+  width <- as.numeric(args[8])
+  height <- as.numeric(args[9])
+  tabfile <- args[10]
+
+  mq_stacked(img, df, all_intcols, json_dump, nterms, target_rank, width, height, tabfile)
+}
 
 ####### ==================== #######
 #              MAIN                #

setup.py

@@ -1,6 +1,6 @@
 from setuptools import setup, find_packages
 
-VERSION = '1.0.0'
+VERSION = '1.1.0'
 URL = 'https://github.com/galaxyproteomics/metaquantome'
 AUTHOR = 'Caleb Easterly'
 AUTHOR_EMAIL = '[email protected]'