Mercurial > repos > proteore > proteore_expression_levels_by_tissue

changeset 12:56c93d1fc8fd draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
author proteore
date Fri, 28 Jun 2019 05:10:08 -0400
parents 6e85f07c20ee
children 2173192d4824
files get_expression_profiles.R get_expression_profiles.xml proteore_protein_atlas_normal_tissue.loc.sample proteore_protein_atlas_tumor_tissue.loc.sample tool_data_table_conf.xml.sample
diffstat 5 files changed, 42 insertions(+), 22 deletions(-) [+]
line wrap: on
line diff
--- a/get_expression_profiles.R	Wed Mar 13 09:30:12 2019 -0400
+++ b/get_expression_profiles.R	Fri Jun 28 05:10:08 2019 -0400
@@ -69,6 +69,15 @@
   return(res)
 }
 
+clean_ids <- function(ids){
+  
+  ids = gsub(" ","",ids)
+  ids = ids[which(ids!="")]
+  ids = ids[which(ids!="NA")]
+  ids = ids[!is.na(ids)]
+ 
+  return(ids) 
+}
 
 main <- function() {
   args <- commandArgs(TRUE)
@@ -109,16 +118,17 @@
   # Extract input
   input_type = args$input_type
   if (input_type == "list") {
-    list_id = strsplit(args$input, "[ \t\n]+")[[1]]
+    ids = unlist(strsplit(strsplit(args$input, "[ \t\n]+")[[1]],";"))
   } else if (input_type == "file") {
     filename = args$input
     column_number = as.numeric(gsub("c", "" ,args$column_number))
     header = str2bool(args$header)
     file = read_file(filename, header)
-    list_id = sapply(strsplit(file[,column_number], ";"), "[", 1)
+    ids = unlist(strsplit(file[,column_number], ";"))
   }
-  input = list_id
-
+  #filter ids
+  ids = clean_ids(ids)
+  
   # Read reference file
   reference_file = read_file(args$ref_file, TRUE)
 
@@ -130,11 +140,11 @@
     level = strsplit(args$level, ",")[[1]]
     reliability = strsplit(args$reliability, ",")[[1]]
     # Calculation
-    res = annot.HPAnorm(input, reference_file, tissue, level, reliability, not_mapped_option)
+    res = annot.HPAnorm(ids, reference_file, tissue, level, reliability, not_mapped_option)
   } else if (atlas=="cancer") {
     cancer = strsplit(args$cancer, ",")[[1]]
     # Calculation
-    res = annot.HPAcancer(input, reference_file, cancer, not_mapped_option)
+    res = annot.HPAcancer(ids, reference_file, cancer, not_mapped_option)
   }
   
   # Write output
--- a/get_expression_profiles.xml	Wed Mar 13 09:30:12 2019 -0400
+++ b/get_expression_profiles.xml	Fri Jun 28 05:10:08 2019 -0400
@@ -1,4 +1,4 @@
-<tool id="sel_ann_hpa" name="Get expression profiles" version="2019.03.13">
+<tool id="sel_ann_hpa" name="Get expression profiles" version="2019.06.27">
     <description>by (normal or tumor) tissue/cell type [Human Protein Atlas]
     </description>
     <requirements>
@@ -38,7 +38,7 @@
     </command>
        <inputs>
            <conditional name="input" >
-            <param name="ids" type="select" label="Enter your IDs (ENSG IDs only)" help="Copy/paste or from a file (e.g. table)" >
+            <param name="ids" type="select" label="Enter your IDs (Ensembl gene ID only; e.g. ENSG00000163631)" help="Copy/paste or from a file (e.g. table)" >
                 <option value="list">Copy/paste your IDs</option>
                 <option value="file" selected="true">Input file containing your IDs</option>
             </param>
@@ -68,8 +68,10 @@
                 <option value="tumor">Human Tumor Tissue</option>
             </param>
             <when value="normal">
-                <param name="proteinatlas" type="select" label="Normal tissue HPA version" >
-                    <options from_data_table="proteore_protein_atlas_normal_tissue"/>
+                <param name="proteinatlas" type="select" label="Normal tissue HPA version 18.1" >
+                    <options from_data_table="proteore_protein_atlas_normal_tissue">
+                        <filter type="sort_by" column="0"/>
+                    </options>
 	            </param>
                 <param name="normal_tissue" type="select" label="Select tissue(s)" multiple="True" display="checkboxes" optional="False">
                     <option value="adrenal gland" >Adrenal gland</option>
@@ -123,7 +125,7 @@
                 </param>
                 <param name="level" type="select" label="Expression level" display="checkboxes" multiple="True" optional="False">
                     <option value="High" selected="true">High</option>
-                    <option value="Medium">Medium</option>
+                    <option value="Medium" selected="true">Medium</option>
                     <option value="Low">Low</option>
                     <option value="Not detected">Not detected</option>
                 </param>
@@ -135,8 +137,10 @@
                 </param>
             </when>
             <when value="tumor">
-                <param name="proteinatlas" type="select" label="Tumor tissue HPA version" >
-                    <options from_data_table="proteore_protein_atlas_tumor_tissue"/>
+                <param name="proteinatlas" type="select" label="Tumor tissue HPA version 18.1" >
+                    <options from_data_table="proteore_protein_atlas_tumor_tissue">
+                        <filter type="sort_by" column="0"/>
+                    </options>
 	            </param>
                 <param name="cancer_tissue" type="select" label="Select cancer tissue(s)" multiple="True" display="checkboxes" optional="False">
                     <option value="breast cancer" >Breast cancer</option>
@@ -194,7 +198,13 @@
  
 **Input**
 
-A list of ENSG (Ensembl gene) IDs must be entered (either via a copy/paste or by choosing a file); if it's not the case, please use the ID_Converter tool of ProteoRE.
+A list of Ensembl gene IDs (e.g. ENSG00000163631) must be entered (either via a copy/paste or by choosing a file); if it's not the case, please use the ID_Converter tool of ProteoRE.
+
+A line can contains multiple ids if separated by ";". 
+
+.. class:: warningmark
+	
+In copy/paste mode, the number of IDs considered in input is limited to 5000.
 
 -----
 
@@ -232,7 +242,7 @@
 
 **Data sources (release date)**
 
-Both normal and tumor tissues data are based on the Human Protein Atlas version 18 and Ensembl version 88.38.
+Both normal and tumor tissues data are based on the Human Protein Atlas version 18.1 and Ensembl version 88.38. More information: https://www.proteinatlas.org/about/download
 
 -----
 
@@ -246,7 +256,7 @@
 
 This work has been partially funded through the French National Agency for Research (ANR) IFB project.
 
-Contact support@proteore.org for any questions or concerns about the Galaxy implementation of this tool.
+Help: contact@proteore.org for any questions or concerns about this tool.
         
     ]]></help>
     <citations>
--- a/proteore_protein_atlas_normal_tissue.loc.sample	Wed Mar 13 09:30:12 2019 -0400
+++ b/proteore_protein_atlas_normal_tissue.loc.sample	Fri Jun 28 05:10:08 2019 -0400
@@ -1,2 +1,2 @@
-#<id>	<name>	<tissue>	<value>
-HPA_normal_tissue_23-10-2018	HPA normal tissue 23/10/2018	HPA_normal_tissue	/tool-data/HPA_normal_tissue_23-10-2018.tsv
+#<id>	<release>	<name>	<tissue>	<value>
+9979818977	HPA_normal_tissue_23-10-2018	HPA normal tissue 23/10/2018	HPA_normal_tissue	/tool-data/HPA_normal_tissue_23-10-2018.tsv
--- a/proteore_protein_atlas_tumor_tissue.loc.sample	Wed Mar 13 09:30:12 2019 -0400
+++ b/proteore_protein_atlas_tumor_tissue.loc.sample	Fri Jun 28 05:10:08 2019 -0400
@@ -1,2 +1,2 @@
-#<id>	<name>	<tissue>	<value>
-HPA_pathology_23-10-2018	HPA pathology 23/10/2018	HPA_pathology	/tool-data/HPA_pathology_23-10-2018.tsv
+#<id>	<release>	<name>	<tissue>	<value>
+9979818977	HPA_pathology_23-10-2018	HPA pathology 23/10/2018	HPA_pathology	/tool-data/HPA_pathology_23-10-2018.tsv
--- a/tool_data_table_conf.xml.sample	Wed Mar 13 09:30:12 2019 -0400
+++ b/tool_data_table_conf.xml.sample	Fri Jun 28 05:10:08 2019 -0400
@@ -1,10 +1,10 @@
 <tables>
     <table name="proteore_protein_atlas_normal_tissue" comment_char="#">
-      <columns>id, name, tissue, value</columns>
+      <columns>id, release, name, tissue, value</columns>
       <file path="tool-data/proteore_protein_atlas_normal_tissue.loc" />
     </table>
     <table name="proteore_protein_atlas_tumor_tissue" comment_char="#">
-      <columns>id, name, tissue, value</columns>
+      <columns>id, release, name, tissue, value</columns>
       <file path="tool-data/proteore_protein_atlas_tumor_tissue.loc" />
     </table>
 </tables>
\ No newline at end of file