Expand the set of allowed species

NEWS.md

@@ -6,6 +6,7 @@
 * Add argument to define the assays(s) to use for exported values and barplots
 * Harmonize treatment of merged groups in runTest and plotVolcano
 * Sort exported test results by p-value instead of logFC
+* Expand the set of allowed species
 
 # einprot 0.7.3
 

R/getSupportedSpecies.R

@@ -14,13 +14,26 @@
 #'
 getSupportedSpecies <- function() {
     data.frame(
-        taxId = c(10090, 9606, 6239, 7955, 7227, 4932, 284812),
+        taxId = c(10090, 9606, 6239, 7955, 7227, 4932, 284812, 28377,
+                  9913, 9615, 9796, 9685, 9031, 9544, 13616, 9258,
+                  9598, 10116, 9823, 8364),
         species = c("Mus musculus", "Homo sapiens", "Caenorhabditis elegans",
                     "Danio rerio", "Drosophila melanogaster",
                     "Saccharomyces cerevisiae",
-                    "Schizosaccharomyces pombe 972h-"),
+                    "Schizosaccharomyces pombe 972h-",
+                    "Anolis carolinensis", "Bos taurus",
+                    "Canis lupus familiaris", "Equus caballus",
+                    "Felis catus", "Gallus gallus",
+                    "Macaca mulatta", "Monodelphis domestica",
+                    "Ornithorhynchus anatinus", "Pan troglodytes",
+                    "Rattus norvegicus", "Sus scrofa",
+                    "Xenopus tropicalis"),
         speciesCommon = c("mouse", "human", "roundworm", "zebrafish",
-                          "fruitfly", "baker's yeast", "fission yeast")
+                          "fruitfly", "baker's yeast", "fission yeast",
+                          "green anole", "bovine", "dog", "horse", "cat",
+                          "chicken", "rhesus macaque", "opossum",
+                          "platypus", "chimpanzee", "Norway rat",
+                          "pig", "tropical clawed frog")
     )
 }
 
@@ -44,6 +57,8 @@ getSupportedSpecies <- function() {
 #' getSpeciesInfo("mouse")
 #' getSpeciesInfo(6239)
 #' getSpeciesInfo("Homo sapiens")
+#' ## unsupported species
+#' getSpeciesInfo("E.coli")
 #'
 getSpeciesInfo <- function(species) {
     stopifnot(length(species) == 1)
@@ -56,21 +71,26 @@ getSpeciesInfo <- function(species) {
         species_common <-
             taxTable$speciesCommon[match(tolower(species),
                                          tolower(taxTable$speciesCommon))]
+        tax_id <- taxTable$taxId[match(species_id, taxTable$species)]
     } else if (tolower(species) %in% tolower(taxTable$species)) {
         species_id <- taxTable$species[match(tolower(species),
                                              tolower(taxTable$species))]
         species_common <-
             taxTable$speciesCommon[match(tolower(species),
                                          tolower(taxTable$species))]
+        tax_id <- taxTable$taxId[match(species_id, taxTable$species)]
     } else if (species %in% taxTable$taxId) {
         species_id <- taxTable$species[match(species,
                                              taxTable$taxId)]
         species_common <- taxTable$speciesCommon[match(species,
                                                        taxTable$taxId)]
+        tax_id <- taxTable$taxId[match(species_id, taxTable$species)]
     } else {
-        stop("Unknown species ", species)
+        warning("Unknown species ", species)
+        species_id <- species
+        species_common <- ""
+        tax_id <- NA_real_
     }
-    tax_id <- taxTable$taxId[match(species_id, taxTable$species)]
 
     list(species = species_id,
          speciesCommon = species_common,

R/getUniProtToGeneSymbolMapping.R

@@ -18,6 +18,10 @@
         f <- "YEAST_559292_idmapping.dat.gz"
     } else if (spi$species == "Schizosaccharomyces pombe 972h-") {
         f <- "SCHPO_284812_idmapping.dat.gz"
+    } else if (spi$species == "Gallus gallus") {
+        f <- "CHICK_9031_idmapping.dat.gz"
+    } else if (spi$species == "Rattus norvegicus") {
+        f <- "RAT_10116_idmapping.dat.gz"
     } else {
         stop("Unsupported species")
     }

R/makeDbLinkTable.R

@@ -213,7 +213,7 @@ makeDbLinkTable <- function(df, idCol, speciesCommon,
     .assertVector(x = df, type = "data.frame")
     .assertScalar(x = idCol, type = "character", validValues = colnames(df))
     .assertScalar(x = speciesCommon, type = "character",
-                  validValues = getSupportedSpecies()$speciesCommon)
+                  validValues = c("", getSupportedSpecies()$speciesCommon))
     .assertScalar(x = addSpeciesSpecificColumns, type = "logical")
     .assertVector(x = convTablePomBase, type = "data.frame", allowNULL = TRUE)
     .assertVector(x = convTableWormBase, type = "data.frame",

R/prepareFeatureCollections.R

@@ -135,21 +135,25 @@ prepareFeatureCollections <- function(sce, idCol, includeFeatureCollections,
     ## -------------------------------------------------------------------------
     if ("complexes" %in% includeFeatureCollections) {
         complexes <- readRDS(complexDbPath)
-        if (speciesInfo$speciesCommon %in% names(complexes)) {
-            crl <- complexes[[speciesInfo$speciesCommon]]
-        } else if (speciesInfo$species %in% names(complexes)) {
-            crl <- complexes[[speciesInfo$species]]
+        if (any(c(speciesInfo$speciesCommon, speciesInfo$species) %in%
+                names(complexes))) {
+            if (speciesInfo$speciesCommon %in% names(complexes)) {
+                crl <- complexes[[speciesInfo$speciesCommon]]
+            } else if (speciesInfo$species %in% names(complexes)) {
+                crl <- complexes[[speciesInfo$species]]
+            }
+            if (complexSpecies == "current") {
+                ## Only test complexes defined for the current species
+                crl <- crl[S4Vectors::mcols(crl)$Species.common %in%
+                               c(speciesInfo$species, speciesInfo$speciesCommon)]
+            }
+            S4Vectors::mcols(crl)$genes <- vapply(
+                crl, function(w) gsub(pat, "\\1; ", paste(w, collapse = ";")), "")
+            S4Vectors::mcols(crl)$nGenes <- lengths(crl)
         } else {
-            stop("No complex database available for the current species")
+            warning("No complex database available for the current species")
+            crl <- IRanges::CharacterList()
         }
-        if (complexSpecies == "current") {
-            ## Only test complexes defined for the current species
-            crl <- crl[S4Vectors::mcols(crl)$Species.common %in%
-                           c(speciesInfo$species, speciesInfo$speciesCommon)]
-        }
-        S4Vectors::mcols(crl)$genes <- vapply(
-            crl, function(w) gsub(pat, "\\1; ", paste(w, collapse = ";")), "")
-        S4Vectors::mcols(crl)$nGenes <- lengths(crl)
     } else {
         crl <- IRanges::CharacterList()
     }
@@ -185,7 +189,8 @@ prepareFeatureCollections <- function(sce, idCol, includeFeatureCollections,
     ## -------------------------------------------------------------------------
     ## GO terms
     ## -------------------------------------------------------------------------
-    if ("GO" %in% includeFeatureCollections) {
+    if ("GO" %in% includeFeatureCollections &&
+        speciesInfo$species %in% getSupportedSpecies()$species) {
         goannots <- msigdbr::msigdbr(species = speciesInfo$species,
                                      category = "C5") %>%
             dplyr::select("gs_name", "gene_symbol")

inst/extdata/process_FragPipe_template.Rmd

@@ -72,8 +72,9 @@ of the data are provided via [principal component analysis](#run-pca) and
 ## Get species info and define STRINGdb object
 speciesInfo <- getSpeciesInfo(species)
 if (is.null(stringDir)) stringDir <- ""
-if (is.null(stringIdCol)) {
-    ## If no STRING IDs are extracted, don't do STRING analysis
+if (is.null(stringIdCol) || is.na(speciesInfo$taxId)) {
+    ## If no STRING IDs are extracted or an unknown species is provided, 
+    ## don't do STRING analysis
     string_db <- NULL
 } else {
     string_db <- tryCatch({

inst/extdata/process_MaxQuant_template.Rmd

@@ -72,8 +72,9 @@ of the data are provided via [principal component analysis](#run-pca) and
 ## Get species info and define STRINGdb object
 speciesInfo <- getSpeciesInfo(species)
 if (is.null(stringDir)) stringDir <- ""
-if (is.null(stringIdCol)) {
-    ## If no STRING IDs are extracted, don't do STRING analysis
+if (is.null(stringIdCol) || is.na(speciesInfo$taxId)) {
+    ## If no STRING IDs are extracted or an unknown species is provided, 
+    ## don't do STRING analysis
     string_db <- NULL
 } else {
     string_db <- tryCatch({

inst/extdata/process_PD_TMT_template.Rmd

@@ -73,8 +73,9 @@ speciesInfo <- getSpeciesInfo(species)
 
 if (inputLevel == "Proteins") {
     if (is.null(stringDir)) stringDir <- ""
-    if (is.null(stringIdCol)) {
-        ## If no STRING IDs are extracted, don't do STRING analysis
+    if (is.null(stringIdCol) || is.na(speciesInfo$taxId)) {
+        ## If no STRING IDs are extracted or an unknown species is provided, 
+        ## don't do STRING analysis
         string_db <- NULL
     } else {
         string_db <- tryCatch({

tests/testthat/test-checkArgumentsFragPipe.R

@@ -149,8 +149,8 @@ test_that("argument checking for FP works", {
     ## species
     args <- args0
     args$species <- 1
-    expect_error(do.call(.checkArgumentsFragPipe, args),
-                 "Unknown species 1")
+    expect_warning(do.call(.checkArgumentsFragPipe, args),
+                   "Unknown species 1")
     args$species <- c("Mouse", "Human")
     expect_error(do.call(.checkArgumentsFragPipe, args),
                  "length(species) == 1 is not TRUE", fixed = TRUE)

tests/testthat/test-checkArgumentsMaxQuant.R

@@ -157,8 +157,8 @@ test_that("argument checking for MQ works", {
     ## species
     args <- args0
     args$species <- 1
-    expect_error(do.call(.checkArgumentsMaxQuant, args),
-                 "Unknown species 1")
+    expect_warning(do.call(.checkArgumentsMaxQuant, args),
+                   "Unknown species 1")
     args$species <- c("Mouse", "Human")
     expect_error(do.call(.checkArgumentsMaxQuant, args),
                  "length(species) == 1 is not TRUE", fixed = TRUE)

tests/testthat/test-checkArgumentsPDTMT.R

@@ -167,8 +167,8 @@ test_that("argument checking for PD-TMT works", {
     ## species
     args <- args0
     args$species <- 1
-    expect_error(do.call(.checkArgumentsPDTMT, args),
-                 "Unknown species 1")
+    expect_warning(do.call(.checkArgumentsPDTMT, args),
+                   "Unknown species 1")
     args$species <- c("Mouse", "Human")
     expect_error(do.call(.checkArgumentsPDTMT, args),
                  "length(species) == 1 is not TRUE", fixed = TRUE)

tests/testthat/test-checkArgumentsPDTMTptm.R

@@ -124,8 +124,8 @@ test_that("argument checking for PD-TMT PTM works", {
     ## species
     args <- args0
     args$species <- 1
-    expect_error(do.call(.checkArgumentsPDTMTptm, args),
-                 "Unknown species 1")
+    expect_warning(do.call(.checkArgumentsPDTMTptm, args),
+                   "Unknown species 1")
     args$species <- c("Mouse", "Human")
     expect_error(do.call(.checkArgumentsPDTMTptm, args),
                  "length(species) == 1 is not TRUE", fixed = TRUE)

tests/testthat/test-getSupportedSpecies.R

@@ -3,10 +3,10 @@ test_that("geting species information works", {
     expect_s3_class(df, "data.frame")
     expect_equal(ncol(df), 3)
     expect_named(df, c("taxId", "species", "speciesCommon"))
-    expect_equal(nrow(df), 7)
+    expect_equal(nrow(df), 20)
 
-    expect_error(getSpeciesInfo("missing"),
-                 "Unknown species missing")
+    expect_warning(getSpeciesInfo("missing"),
+                   "Unknown species missing")
     expect_error(getSpeciesInfo(list(x = 1)),
                  "is.character(species) || is.numeric(species) is not TRUE",
                  fixed = TRUE)

tests/testthat/test-getUniProtToGeneSymbolMapping.R

@@ -1,6 +1,7 @@
 test_that("getUniProtToIDMapping works", {
-    expect_error(getUniProtToIDMapping(1), "Unknown species")
-    expect_error(getUniProtToIDMapping("wrongSpecies"), "Unknown species")
+    expect_warning(expect_error(getUniProtToIDMapping(1), "Unsupported species"))
+    expect_warning(expect_error(getUniProtToIDMapping("wrongSpecies"),
+                                "Unsupported species"))
     expect_error(getUniProtToIDMapping("fruitfly", targetId = 1),
                  "'targetId' must be of class 'character'")
 
@@ -22,6 +23,10 @@ test_that("getUniProtToIDMapping works", {
                  "YEAST_559292_idmapping.dat.gz")
     expect_equal(.getUniProtToIDMappingFile(getSpeciesInfo("fission yeast")),
                  "SCHPO_284812_idmapping.dat.gz")
+    expect_equal(.getUniProtToIDMappingFile(getSpeciesInfo("Norway rat")),
+                 "RAT_10116_idmapping.dat.gz")
+    expect_equal(.getUniProtToIDMappingFile(getSpeciesInfo("chicken")),
+                 "CHICK_9031_idmapping.dat.gz")
     expect_error(.getUniProtToIDMappingFile(list(species = "missing")),
                  "Unsupported species")
 

tests/testthat/test-prepareFeatureCollections.R

@@ -177,8 +177,8 @@ test_that("preparing feature collections works", {
     expect_error(do.call(prepareFeatureCollections, args),
                  "All values in 'namesspeciesInfo' must be one of")
     args$speciesInfo <- list(speciesCommon = "missing", species = "missing")
-    expect_error(do.call(prepareFeatureCollections, args),
-                 "No complex database available for the current species")
+    expect_warning(do.call(prepareFeatureCollections, args),
+                   "No complex database available for the current species")
 
     args <- args0
     args$complexSpecies <- 1

tests/testthat/test-runFragPipeAnalysis.R

@@ -145,9 +145,6 @@ test_that("runFragPipeAnalysis works", {
 
     ## species
     args <- args0
-    args$species <- 1
-    expect_error(do.call(runFragPipeAnalysis, args),
-                 "Unknown species 1")
     args$species <- c("Mouse", "Human")
     expect_error(do.call(runFragPipeAnalysis, args),
                  "length(species) == 1 is not TRUE", fixed = TRUE)

tests/testthat/test-runMaxQuantAnalysis.R

@@ -161,9 +161,6 @@ test_that("runMaxQuantAnalysis works", {
 
     ## species
     args <- args0
-    args$species <- 1
-    expect_error(do.call(runMaxQuantAnalysis, args),
-                 "Unknown species 1")
     args$species <- c("Mouse", "Human")
     expect_error(do.call(runMaxQuantAnalysis, args),
                  "length(species) == 1 is not TRUE", fixed = TRUE)

tests/testthat/test-runPDTMTAnalysis.R

@@ -181,9 +181,6 @@ test_that("runPDTMTAnalysis works", {
 
     ## species
     args <- args0
-    args$species <- 1
-    expect_error(do.call(runPDTMTAnalysis, args),
-                 "Unknown species 1")
     args$species <- c("Mouse", "Human")
     expect_error(do.call(runPDTMTAnalysis, args),
                  "length(species) == 1 is not TRUE", fixed = TRUE)

tests/testthat/test-runPDTMTptmAnalysis.R

@@ -122,9 +122,6 @@ test_that("runPDTMTptmAnalysis works", {
 
     ## species
     args <- args0
-    args$species <- 1
-    expect_error(do.call(runPDTMTptmAnalysis, args),
-                 "Unknown species 1")
     args$species <- c("Mouse", "Human")
     expect_error(do.call(runPDTMTptmAnalysis, args),
                  "length(species) == 1 is not TRUE", fixed = TRUE)