Merge pull request #36 from senresearch/dev

update to v0.3.0

Project.toml

@@ -1,7 +1,7 @@
 name = "GeneNetworkAPI"
 uuid = "5ef98721-7938-4c15-bad2-60f0fa16d62a"
 authors = ["Chelsea Trotter <chelsea.xhu@gmail.com>", "Gregory Farage <gfarage@uthsc.edu>", "Saunak Sen <sen@uthsc.edu>"]
-version = "0.2.0"
+version = "0.3.0"
 
 [deps]
 CSV = "336ed68f-0bac-5ca0-87d4-7b16caf5d00b"

README.md

@@ -74,7 +74,7 @@ julia> GeneNetworkAPI.subset(list_species(), :Name => x->x.=="rat")
 ## List groups for a species
 
 Since the information is organized by segregating population
-("group"), it is useful to get a list for a preticular species you
+("group"), it is useful to get a list for a particular species you
 might be interested in.
 
 ```

src/GeneNetworkAPI.jl

@@ -13,6 +13,9 @@ module GeneNetworkAPI
     include("./info_data.jl")
     export info_dataset, info_pheno
 
+    include("./download_data.jl")
+    export download_geno, download_pheno, download_omics
+
     include("./get_data.jl")
     export get_geno, get_pheno, get_omics
 
@@ -23,7 +26,7 @@ module GeneNetworkAPI
     export show_table
 
     include("./utils_geno.jl")
-    export parse_geno, genofile_location
+    export parse_geno, genofile_location, has_genofile_meta
 
     include("./utils.jl")
     export make_rectangular, has_error_500

src/download_data.jl

@@ -0,0 +1,90 @@
+# This file contains the functions to download data from gene network APIs
+
+##############################
+# Functions downloading data #
+##############################
+
+#############
+# Geno data #
+#############
+"""                                                                                    
+    download_geno(group::String,format::String="geno";
+                    gn_url::String=gn_url(),
+                    path::String = "")
+
+Download the genotype matrix for a `group` in a given `format`.
+
+Currently works only for files in the `geno` format.
+
+The file will be downloaded in `path` location, if  `path` is not specified, then 
+the file will be a temporary file.
+
+"""
+function download_geno(group, format = "geno";
+                        gn_url::String = gn_url(),
+                        path::String = tempname())
+
+	if has_genofile_meta(group; gn_url = gn_url)
+		# need to check real location of data
+		vlocation = genofile_location(group; gn_url = gn_url)
+
+		if length(vlocation) == 1
+			group = vlocation[1][1:end-5] # expect ".geno" extension
+			# else
+			# println("Info: group ", group, " has additional genotype files, see locations")
+			# show_table(list_geno(group; gn_url=gn_url))
+		end
+	end
+
+	# increase timeout response from the server
+	downloader = Downloads.Downloader()
+	downloader.easy_hook = (easy, info) -> Downloads.Curl.setopt(easy, Downloads.Curl.CURLOPT_LOW_SPEED_TIME, 300)
+
+	geno_url = gn_url * "genotypes/" * group * "." * format
+
+	return Downloads.download(geno_url, path; downloader = downloader)
+end
+
+##############
+# Pheno data #
+##############
+
+"""                                                                                    
+    download_pheno(dataset::String;gn_url::String=gn_url())
+
+Downloads the non-omic ("clinical") phenotypes for a given `dataset`.
+The file will be downloaded in `path` location, if  `path` is not specified, then 
+the file will be a temporary file.
+"""
+function download_pheno(dataset::String; 
+                        gn_url::String=gn_url(),
+                        path::String = tempname())
+    pheno_url = gn_url * "sample_data" * "/" * dataset * "Publish"
+    return Downloads.download(pheno_url, path)
+end
+
+##############
+# Omics data #
+##############
+
+
+"""                                                                                    
+    download_omics(dataset::String; 
+                    gn_url::String=gn_url(), 
+                    path::String = tempname())
+
+Download the omic phenotypes for a given `dataset`.
+The file will be downloaded in `path` location, if  `path` is not specified, then 
+the file will be a temporary file.
+"""
+function download_omics(dataset::String; 
+                        gn_url::String=gn_url(), 
+                        path::String = tempname())
+# increase timeout response from the server
+    downloader = Downloads.Downloader();
+    downloader.easy_hook = (easy, info) -> Downloads.Curl.setopt(easy, Downloads.Curl.CURLOPT_LOW_SPEED_TIME, 300);
+
+    omics_url = gn_url * "sample_data" * "/" * dataset
+
+    return Downloads.download(omics_url, path; downloader=downloader)
+end

src/get_data.jl

@@ -17,24 +17,9 @@ Currently works only for files in the `geno` format.
 """
 function get_geno(group, format="geno"; gn_url::String=gn_url())
 
-    if has_genofile_meta(group; gn_url=gn_url)
-        # need to check real location of data
-        vlocation = genofile_location(group; gn_url=gn_url)
-
-        if length(vlocation) == 1 
-            group = vlocation[1][1:end-5] # expect ".geno" extension
-        # else
-            # println("Info: group ", group, " has additional genotype files, see locations")
-            # show_table(list_geno(group; gn_url=gn_url))
-        end
-    end
-
-    # increase timeout response from the server
-    downloader = Downloads.Downloader();
-    downloader.easy_hook = (easy, info) -> Downloads.Curl.setopt(easy, Downloads.Curl.CURLOPT_LOW_SPEED_TIME, 300);
-
-    geno_url = gn_url * "genotypes/" * group * "." * format
-    geno = parse_geno(Downloads.download(geno_url; downloader=downloader))
+    geno_file = download_geno(group, format;gn_url = gn_url)
+
+    geno = parse_geno(geno_file)
 
     return geno
 end
@@ -47,17 +32,18 @@ end
     get_pheno(dataset::String;gn_url::String=gn_url())
 
 Returns the non-omic ("clinical") phenotypes for a given `dataset`.
-"""
-function get_pheno(dataset::String; gn_url::String=gn_url())
-    url = gn_url * "sample_data" * "/" * dataset * "Publish"
-    return CSV.read(Downloads.download(url), DataFrame, delim=',',missingstring="x")
-end
 
-"""                                                                                    
-    get_pheno(group::String,trait::String="geno";gn_url::String=gn_url())
+---
+
+    get_pheno(dataset::String,trait::String; gn_url::String=gn_url())
 
 Returns a given `trait` in a `group`.
 """
+function get_pheno(dataset::String; gn_url::String=gn_url())
+    pheno_file = download_pheno(dataset; gn_url = gn_url)   
+    return CSV.read(pheno_file, DataFrame, delim=',',missingstring="x")
+end
+
 function get_pheno(dataset::String,trait::String; gn_url::String=gn_url())
     url = gn_url * "sample_data" * "/" * dataset * "/" * trait 
     return json2df(get_api(url))
@@ -68,18 +54,12 @@ end
 # Omics data #
 ##############
 
-
 """                                                                                    
     get_omics(dataset::String;gn_url::String=gn_url())
 
 Returns the omic phenotypes for a given `dataset`.
 """
 function get_omics(dataset::String; gn_url::String=gn_url())
-# increase timeout response from the server
-    downloader = Downloads.Downloader();
-    downloader.easy_hook = (easy, info) -> Downloads.Curl.setopt(easy, Downloads.Curl.CURLOPT_LOW_SPEED_TIME, 300);
-
-    omics_url = gn_url * "sample_data" * "/" * dataset
-
-    return CSV.read(Downloads.download(omics_url; downloader=downloader), DataFrame, delim=',',missingstring="x")
+    omics_file = download_omics(dataset; gn_url= gn_url)
+    return CSV.read(omics_file, DataFrame, delim=',',missingstring="x")
 end

test/gnAPI_tests.jl

@@ -115,3 +115,33 @@ dfRslt1 = GeneNetworkAPI.select(filter(row->row.trait == 12762, dfRslt), [2,3,4]
 
 println("Get correlation test 10: ", @test (dfRslt1 == dfCorrelation));
 
+#########################
+# TEST 11 download geno #
+#########################
+
+file_downloaded = download_geno("BXD", path = "BXD_raw.geno");
+
+println("Download geno file test 11: ", @test (isfile(file_downloaded)));
+
+rm(file_downloaded)
+
+
+##########################
+# TEST 12 download pheno #
+##########################
+
+file_downloaded = download_pheno("HSNIH-Palmer", path = "HSNIH-Palmer.csv");
+
+println("Download pheno file test 12: ", @test (isfile(file_downloaded)));
+
+rm(file_downloaded)
+
+##########################
+# TEST 13 download omics #
+##########################
+
+file_downloaded = download_omics("INIA_AmgCoh_0311");
+
+println("Download pheno file test 13: ", @test (isfile(file_downloaded)));
+
+rm(file_downloaded)