allow maxquant data containing the 'combined' subfolder (#129)

* re-enable maxquant data containing the combined subfolder

Co-authored-by: Steffengreiner <[email protected]>

src/main/groovy/life/qbic/utils/MaxQuantParser.groovy

@@ -118,7 +118,7 @@ class MaxQuantParser implements DatasetParser<MaxQuantRunResult> {
         rootChildren.each { currentChild ->
             if (currentChild.containsKey("children")) {
                 //folder
-                parseTxtFolder(map)
+                parseSubFolders(map)
             } else if (currentChild.containsKey("fileType")) {
                 //file
                 String name = currentChild.get("name")
@@ -132,18 +132,26 @@ class MaxQuantParser implements DatasetParser<MaxQuantRunResult> {
     }
 
     /**
-     * Method which adapts the parsed content of the txt directory in place to the expected file structure.
+     * Method which adapts the parsed content of the txt directory in place to the expected file structure. This directory can be inside an optional 'combined' directory
      * @see {<a href="https://github.com/qbicsoftware/data-model-lib/blob/master/src/test/resources/examples/resultset/maxquant/valid-resultset-example.json">valid datastructure example</a>}
      *
      * After parsing, the files of the txt directory are contained in the children property of the root directory.
      * The underlying datastructure however expects a mapping of the expected files as a Map entry in the root directory.
      * @param maxQuantInformation a nested map representing the parsed fileTree structure
      * @since 1.9.0
      */
-    private static void parseTxtFolder(Map maxQuantInformation) {
+    private static void parseSubFolders(Map maxQuantInformation) {
         List<Map> rootFolderInformation = maxQuantInformation.get("children") as List<Map>
         def txtFolderInformation
         rootFolderInformation.findAll { map ->
+            if (map.get("name") == "combined") {
+                def combinedFolderInformation = map.get("children")  as List<Map>
+                combinedFolderInformation.findAll() { innerMap ->
+                    if (innerMap.get("name") == "txt") {
+                        txtFolderInformation = innerMap.get("children") as List<Map>
+                    }
+                }
+            }
             if (map.get("name") == "txt") {
                 txtFolderInformation = map.get("children")  as List<Map>
             }

src/test/groovy/life/qbic/utils/MaxQuantParserSpec.groovy

@@ -52,6 +52,37 @@ class MaxQuantParserSpec extends Specification {
         assert maxQuantRunResult.proteinGroups.getName()== "proteinGroups.txt"
     }
 
+    def "parsing the old file structure with combined folder returns a maxQuantRunResult object"() {
+        given: "A valid maxQuant run output data structure"
+        def pathToDirectory = Paths.get(exampleDirectoriesRoot, "validates2")
+        when: "we parse this valid structure"
+        MaxQuantRunResult maxQuantRunResult = maxQuantParser.parseFrom(pathToDirectory)
+        then: "we expect no exception should be thrown"
+        assert maxQuantRunResult instanceof MaxQuantRunResult
+        //Root files can be parsed
+        assert maxQuantRunResult.runParameters.getRelativePath() == "./mqpar.xml"
+        assert maxQuantRunResult.runParameters.getName()== "mqpar.xml"
+
+        assert maxQuantRunResult.sampleIds.getRelativePath() == "./QABCD_sample_ids.txt"
+        assert maxQuantRunResult.sampleIds.getName()== "QABCD_sample_ids.txt"
+
+        //Files in ./txt/ can be parsed
+        assert maxQuantRunResult.allPeptides.getRelativePath() == "./combined/txt/allPeptides.txt"
+        assert maxQuantRunResult.allPeptides.getName()== "allPeptides.txt"
+
+        assert maxQuantRunResult.evidence.getRelativePath() == "./combined/txt/evidence.txt"
+        assert maxQuantRunResult.evidence.getName()== "evidence.txt"
+
+        assert maxQuantRunResult.parameters.getRelativePath() == "./combined/txt/parameters.txt"
+        assert maxQuantRunResult.parameters.getName()== "parameters.txt"
+
+        assert maxQuantRunResult.peptides.getRelativePath() == "./combined/txt/peptides.txt"
+        assert maxQuantRunResult.peptides.getName()== "peptides.txt"
+
+        assert maxQuantRunResult.proteinGroups.getRelativePath() == "./combined/txt/proteinGroups.txt"
+        assert maxQuantRunResult.proteinGroups.getName()== "proteinGroups.txt"
+    }
+
     def "parsing an invalid file structure throws DatasetValidationException"() {
         given:
         def pathToDirectory = Paths.get(exampleDirectoriesRoot, "fails/missing_txt_directory")