qbicsoftware · Steffengreiner · Nov 3, 2023 · Nov 2, 2023 · Nov 3, 2023 · Nov 3, 2023
@@ -128,7 +128,7 @@
     <dependency>
       <artifactId>data-model-lib</artifactId>
       <groupId>life.qbic</groupId>
-      <version>2.25.0</version>
+      <version>2.27.0</version>
     </dependency>
     <dependency>
       <groupId>org.mockito</groupId>

@@ -3,6 +3,7 @@ package life.qbic.utils
 import com.fasterxml.jackson.databind.ObjectMapper
 import groovy.json.JsonSlurper
 import groovy.util.logging.Log4j2
+import life.qbic.datamodel.instruments.OxfordNanoporeInstrumentOutputDoradoMinimal
 import life.qbic.datamodel.instruments.OxfordNanoporeInstrumentOutputMinimal
 import net.jimblackler.jsonschemafriend.Schema
 import net.jimblackler.jsonschemafriend.SchemaStore
@@ -14,7 +15,6 @@ import java.nio.file.Path
 import java.nio.file.Paths
 import java.text.ParseException
 import life.qbic.datamodel.datasets.OxfordNanoporeExperiment
-
 import java.util.stream.Collectors
 
 @Log4j2
@@ -95,8 +95,8 @@ class NanoporeParser {
                 jsonStarted = true
             }
             if (jsonStarted) {
-                def split = line.replaceAll("\\s+","").split(":")
-                if(split.size() == 2 && split[1].replaceAll('"',"").size() <= 1){
+                def split = line.replaceAll("\\s+", "").split(":")
+                if (split.size() == 2 && split[1].replaceAll('"', "").size() <= 1) {
                     log.info("Metadata value ${split[0]} missing in ${reportFile["path"]}")
                 }
                 buffer.append(line)
@@ -110,12 +110,11 @@ class NanoporeParser {
         new File(Paths.get(root.toString(), summaryFile["path"].toString()) as String)
                 .readLines().each { line ->
             def split = line.split("=")
-            if(split.size() > 1){
+            if (split.size() > 1) {
                 finalMetaData[split[0]] = split[1]
-            }
-            else {
+            } else {
                 log.info("Metadata value ${split[0]} missing in ${summaryFile["path"]}, defaulting to empty value")
-                finalMetaData[split[0]] =  ""
+                finalMetaData[split[0]] = ""
             }
         }
         return finalMetaData
@@ -178,18 +177,27 @@ class NanoporeParser {
      * @throws net.jimblackler.jsonschemafriend.ValidationException
      */
     private static void validateJson(String json) throws ValidationException {
-        // Step 1: load schema
+        // Step 1: load json
         ObjectMapper objectMapper = new ObjectMapper()
         Object jsonObject = objectMapper.readValue(json, Object)
+
         SchemaStore schemaStore = new SchemaStore()
-        Schema schema = schemaStore.loadSchema(OxfordNanoporeInstrumentOutputMinimal.getSchemaAsStream())
         Validator validator = new Validator()
-        validator.validate(schema, jsonObject)
+        try {
+            //Validate against Fast5 Based Oxford Measurement
+            Schema schema = schemaStore.loadSchema(OxfordNanoporeInstrumentOutputMinimal.getSchemaAsStream())
+            validator.validate(schema, jsonObject)
+        } catch (ValidationException ignored) {
+            //Validate against Pod5 Based Oxford Measurement
+            Schema schema = schemaStore.loadSchema(OxfordNanoporeInstrumentOutputDoradoMinimal.getSchemaAsStream())
+            validator.validate(schema, jsonObject)
+        }
     }
 
     /*
      * Converts a file tree into a json object.
      */
+
     private static class DirectoryConverter {
         private static final PREDEFINED_EXTENSIONS = ["fastq.gz"]
         private static final IGNORED_FOLDERNAMES = ["qc"]
@@ -239,11 +247,11 @@ class NanoporeParser {
             List<File> children = currentDirectory.listFiles()
 
             List<File> visibleChildren = children.stream()
-                    .filter(file -> !file.isHidden()).collect(Collectors.toList());
+                    .filter(file -> !file.isHidden()).collect(Collectors.toList())
 
             for (File file : children) {
                 if (!visibleChildren.contains(file)) {
-                    hiddenFiles.add(file);
+                    hiddenFiles.add(file)
                 }
             }
 
@@ -252,11 +260,11 @@ class NanoporeParser {
                 return !IGNORED_FOLDERNAMES.contains(currentFolderName)
             }.collect {
                 file ->
-                        if (file.isFile()) {
-                            convertFile(file.toPath())
-                        } else if (file.isDirectory()) {
-                            convertDirectory(file.toPath())
-                        }
+                    if (file.isFile()) {
+                        convertFile(file.toPath())
+                    } else if (file.isDirectory()) {
+                        convertDirectory(file.toPath())
+                    }
             }
 
             def convertedDirectory = [

@@ -125,6 +125,41 @@ class NanoporeParserSpec extends Specification {
     thrown(ValidationException)
   }
 
+  def "parsing a valid minimal file structure for dorado based basecalling containing additional unknown files and folder still returns an OxfordNanoporeExperiment Object"() {
+    given:
+    def pathToDirectory = Paths.get(exampleDirectoriesRoot, "validates/QABCD001AB_E12A345a01_PAE12345_nanopore_valid_dorado_minimal")
+    when:
+    def experiment = NanoporeParser.parseFileStructure(pathToDirectory)
+    then:
+    assert experiment instanceof OxfordNanoporeExperiment
+    // Check that the metadata from the report file has been retrieved
+    assert experiment.getMeasurements().get(0).getMachineHost() == "PCT0094"
+    // Check that the metadata from the summary file has been retrieved
+    assert experiment.getMeasurements().get(0).getLibraryPreparationKit() == "SQK-LSK109-XL"
+  }
+
+  def "parsing a valid file structure for dorado based basecalling containing additional unknown files and folder still returns an OxfordNanoporeExperiment Object"() {
+    given:
+    def pathToDirectory = Paths.get(exampleDirectoriesRoot, "validates/QABCD001AB_E12A345a01_PAE12345_nanopore_valid_dorado_example")
+    when:
+    def experiment = NanoporeParser.parseFileStructure(pathToDirectory)
+    then:
+    assert experiment instanceof OxfordNanoporeExperiment
+    // Check that the metadata from the report file has been retrieved
+    assert experiment.getMeasurements().get(0).getMachineHost() == "PCT0094"
+    // Check that the metadata from the summary file has been retrieved
+    assert experiment.getMeasurements().get(0).getLibraryPreparationKit() == "SQK-LSK109-XL"
+  }
+
+  def "parsing an invalid minimal file structure for dorado based basecalling leads to a ValidationException"() {
+    given:
+    def pathToDirectory = Paths.get(exampleDirectoriesRoot, "fails/QABCD001AB_E12A345a01_PAE12345_missing_skip_folder")
+    when:
+    def experiment = NanoporeParser.parseFileStructure(pathToDirectory)
+    then:
+    thrown(ValidationException)
+  }
+
   def "parsing the alternative valid file structure with metadata missing returns an OxfordNanoporeExperiment Object"() {
     given:
     def pathToDirectory = Paths.get(exampleDirectoriesRoot, "validates/QABCD001AB_E12A345a01_PAE12345_nanopore_new_minimal")

@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1,17 @@
+instrument=PCT0094
+position=1-A3-D3
+flow_cell_id=PAE24142
+sample_id=QNANO027AE_E19D023a01_200211
+protocol_group_id=20200211_QNANO
+protocol=sequencing/sequencing_PRO002_DNA:FLO-PRO002:SQK-LSK109-XL
+protocol_run_id=5a7cfc2a-81b0-412d-baa0-51b939cd8e76
+acquisition_run_id=c6028297dff19d01e7c5fba6487de807d1e99c05
+started=2020-02-11T15:52:10.465982+01:00
+acquisition_stopped=2020-02-14T08:39:54.688916+01:00
+processing_stopped=2020-02-14T08:39:58.804639+01:00
+basecalling_enabled=1
+sequencing_summary_file=sequencing_summary_PAE24142_c6028297.txt
+fast5_files_in_final_dest=2189
+fast5_files_in_fallback=0
+fastq_files_in_final_dest=2189
+fastq_files_in_fallback=0
@@ -0,0 +1,52 @@
+Tracking ID
+===========
+
+{
+    "asic_id": "0004A30B0022C63E",
+    "asic_id_eeprom": "0004A30B0022C63E",
+    "asic_temp": "32.631687",
+    "asic_version": "Unknown",
+    "auto_update": "0",
+    "auto_update_source": "https://mirror.oxfordnanoportal.com/software/MinKNOW/",
+    "bream_is_standard": "0",
+    "configuration_version": "1.0.7",
+    "device_id": "1-E9-H9",
+    "device_type": "promethion",
+    "distribution_status": "stable",
+    "distribution_version": "19.12.5",
+    "exp_script_name": "N/A",
+    "exp_script_purpose": "sequencing_run",
+    "exp_start_time": "2020-01-28T15:17:38Z",
+    "flow_cell_id": "PAE26989",
+    "flow_cell_product_code": "FLO-PRO002",
+    "guppy_version": "3.2.8+bd67289",
+    "heatsink_temp": "36.179111",
+    "hostname": "PCT0094",
+    "hublett_board_id": "0132136faade2e15",
+    "hublett_firmware_version": "2.0.12",
+    "installation_type": "nc",
+    "ip_address": "",
+    "local_firmware_file": "1",
+    "mac_address": "",
+    "operating_system": "ubuntu 16.04",
+    "protocol_group_id": "20200128_QNANO",
+    "protocol_run_id": "",
+    "protocols_version": "4.3.16",
+    "run_id": "db9e9383d44d80bbe1e2600c7a7419056610d46d",
+    "sample_id": "QNANO036AD_E19D023b04",
+    "satellite_board_id": "0000000000000000",
+    "satellite_firmware_version": "2.0.12",
+    "usb_config": "firm_1.2.3_ware#rbt_4.5.6_rbt#ctrl#USB3",
+    "version": "3.6.1"
+}
+
+Duty Time
+=========
+
+ID: db9e9383d44d80bbe1e2600c7a7419056610d46d
+
+Channel State,Experiment Time (minutes),State Time (samples),
+strand,0,144832342
+strand,1,158421270
+strand,2,378095352
+strand,3,472685319
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1,17 @@
+instrument=PCT0094
+position=1-A3-D3
+flow_cell_id=PAE24142
+sample_id=QNANO027AE_E19D023a01_200211
+protocol_group_id=20200211_QNANO
+protocol=sequencing/sequencing_PRO002_DNA:FLO-PRO002:SQK-LSK109-XL
+protocol_run_id=5a7cfc2a-81b0-412d-baa0-51b939cd8e76
+acquisition_run_id=c6028297dff19d01e7c5fba6487de807d1e99c05
+started=2020-02-11T15:52:10.465982+01:00
+acquisition_stopped=2020-02-14T08:39:54.688916+01:00
+processing_stopped=2020-02-14T08:39:58.804639+01:00
+basecalling_enabled=1
+sequencing_summary_file=sequencing_summary_PAE24142_c6028297.txt
+fast5_files_in_final_dest=2189
+fast5_files_in_fallback=0
+fastq_files_in_final_dest=2189
+fastq_files_in_fallback=0
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1,52 @@
+Tracking ID
+===========
+
+{
+    "asic_id": "0004A30B0022C63E",
+    "asic_id_eeprom": "0004A30B0022C63E",
+    "asic_temp": "32.631687",
+    "asic_version": "Unknown",
+    "auto_update": "0",
+    "auto_update_source": "https://mirror.oxfordnanoportal.com/software/MinKNOW/",
+    "bream_is_standard": "0",
+    "configuration_version": "1.0.7",
+    "device_id": "1-E9-H9",
+    "device_type": "promethion",
+    "distribution_status": "stable",
+    "distribution_version": "19.12.5",
+    "exp_script_name": "N/A",
+    "exp_script_purpose": "sequencing_run",
+    "exp_start_time": "2020-01-28T15:17:38Z",
+    "flow_cell_id": "PAE26989",
+    "flow_cell_product_code": "FLO-PRO002",
+    "guppy_version": "3.2.8+bd67289",
+    "heatsink_temp": "36.179111",
+    "hostname": "PCT0094",
+    "hublett_board_id": "0132136faade2e15",
+    "hublett_firmware_version": "2.0.12",
+    "installation_type": "nc",
+    "ip_address": "",
+    "local_firmware_file": "1",
+    "mac_address": "",
+    "operating_system": "ubuntu 16.04",
+    "protocol_group_id": "20200128_QNANO",
+    "protocol_run_id": "",
+    "protocols_version": "4.3.16",
+    "run_id": "db9e9383d44d80bbe1e2600c7a7419056610d46d",
+    "sample_id": "QNANO036AD_E19D023b04",
+    "satellite_board_id": "0000000000000000",
+    "satellite_firmware_version": "2.0.12",
+    "usb_config": "firm_1.2.3_ware#rbt_4.5.6_rbt#ctrl#USB3",
+    "version": "3.6.1"
+}
+
+Duty Time
+=========
+
+ID: db9e9383d44d80bbe1e2600c7a7419056610d46d
+
+Channel State,Experiment Time (minutes),State Time (samples),
+strand,0,144832342
+strand,1,158421270
+strand,2,378095352
+strand,3,472685319
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text
@@ -0,0 +1 @@
+This is some text