Merge pull request #6625 from MaraBesemer/main

add: implement tool to add rank names to phyloseq object

Author: bgruening

tools/phyloseq/add_rank_names_to_phyloseq.R

@@ -0,0 +1,71 @@
+#!/usr/bin/env Rscript
+
+suppressPackageStartupMessages(library("optparse"))
+suppressPackageStartupMessages(library("phyloseq"))
+suppressPackageStartupMessages(library("tidyverse"))
+
+# Option parsing
+option_list <- list(
+    make_option(c("--input"),
+        action = "store", dest = "input",
+        help = "Input file containing a phyloseq object"
+    ),
+    make_option(c("--output"),
+        action = "store", dest = "output",
+        help = "Output file for the updated phyloseq object"
+    ),
+    make_option(c("--ranks"),
+        action = "store", dest = "ranks",
+        help = "Comma-separated list of taxonomy ranks (default: Kingdom,Phylum,Class,Order,Family,Genus,Species)"
+    )
+)
+
+parser <- OptionParser(usage = "%prog [options] file", option_list = option_list)
+args <- parse_args(parser, positional_arguments = TRUE)
+opt <- args$options
+
+cat("Input file: ", opt$input, "\n")
+cat("Output file: ", opt$output, "\n")
+cat("Ranks provided: ", opt$ranks, "\n")
+
+if (is.null(opt$ranks)) {
+    opt$ranks <- "Kingdom,Phylum,Class,Order,Family,Genus,Species"
+}
+
+# Parse rank names
+rank_names <- unlist(strsplit(opt$ranks, ","))
+
+# Load phyloseq object
+physeq <- readRDS(opt$input)
+
+# Check if physeq object is loaded successfully
+if (is.null(physeq)) {
+    stop("Error: Failed to load the Phyloseq object. Check the input file.")
+}
+
+cat("Phyloseq object successfully loaded.\n")
+cat("Class of loaded object: ", class(physeq), "\n")
+
+# Check the current tax_table
+cat("Current tax_table:\n")
+print(tax_table(physeq))
+
+
+# Strict check for taxonomy table and provided ranks
+if (ncol(tax_table(physeq)) != length(rank_names)) {
+    stop(
+        "Error: Number of columns in tax_table does not match the number of provided ranks. ",
+        "Please ensure the taxonomy table matches the ranks exactly."
+    )
+}
+
+# Set column names to the provided ranks
+colnames(tax_table(physeq)) <- rank_names
+
+# Confirm the changes
+cat("Updated tax_table:\n")
+print(tax_table(physeq))
+
+# Save the updated phyloseq object
+saveRDS(physeq, file = opt$output, compress = TRUE)
+cat("Updated Phyloseq object saved to: ", opt$output, "\n")

tools/phyloseq/add_rank_names_to_phyloseq.xml

@@ -0,0 +1,37 @@
+<tool id="phyloseq_add_rank_names" name="Add Rank Names to Phyloseq Object" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
+    <description>Add taxonomy rank names to a phyloseq object</description> 
+    <macros>
+        <import>macros.xml</import>
+    </macros>
+    <expand macro="bio_tools"/>
+    <expand macro="requirements"/>
+    <command detect_errors="exit_code"><![CDATA[
+Rscript '${__tool_directory__}/add_rank_names_to_phyloseq.R' --input '$input'  --output '$output' --ranks '$ranks'
+]]></command>
+
+    <inputs>
+        <expand macro="phyloseq_input"/>
+        <param name="ranks" type="text" label="Comma-separated list of taxonomy ranks"  value="Kingdom,Phylum,Class,Order,Family,Genus,Species"/>
+    </inputs>
+    
+    <outputs>
+        <data name="output" format="phyloseq"/>
+    </outputs>
+
+     <tests>
+        <test>
+            <param name="input" value="output.phyloseq" ftype="phyloseq"/>
+            <param name="ranks" value="Kingdom,Phylum,Class,Order,Family,Genus"/>
+            <output name="output" ftype="phyloseq">
+                <assert_contents>
+                    <has_size value="87125" delta="1000"/>
+                </assert_contents>
+            </output>
+        </test>
+    </tests>
+
+    <help>
+    This tool adds taxonomy rank names to a phyloseq object in the `tax_table` slot.
+    </help>
+    <expand macro="citations"/>
+</tool>