Release/2.21.0 (#331)

* increase snapshot to 2.20.0

* Adjust Nf-core pipeline result to account for sarekv3 file structure

* Remove replacement of hyphens with underscores in project spaces (#326)

* add digiwest property tpes

* Support nanopore data structure v3 (#328)

* add new version of valid nanopore structure

* implement additional v3 changes

* Update schema and test to current file structure

Co-authored-by: Steffengreiner <Steffen.Greiner@gmx.de>

* Fix Nanopore Schema V3 naming and streamline children of other report definition

* Adapt schema and add test case for mibi datastructure

Co-authored-by: wow-such-code <andreas.friedrich@uni-tuebingen.de>

Author: Steffengreiner

src/main/groovy/life/qbic/datamodel/datasets/OxfordNanoporeExperiment.groovy

@@ -33,9 +33,14 @@ final class OxfordNanoporeExperiment implements ExperimentFolder {
             FQDN_FILES + ".MuxScanDataLog",
             FQDN_FILES + ".ReportMdLog",
             FQDN_FILES + ".ReportPDFLog",
+            FQDN_FILES + ".ReportHTMLLog",
+            FQDN_FILES + ".ReportJSONLog",
             FQDN_FILES + ".SequencingSummaryLog",
             FQDN_FILES + ".ThroughputLog",
-            FQDN_FILES + ".BarcodeAlignmentLog"
+            FQDN_FILES + ".BarcodeAlignmentLog",
+            FQDN_FILES + ".PoreActivityLog",
+            FQDN_FILES + ".SampleSheetLog",
+            FQDN_FILES + ".PoreScanDataLog"
     ]
 
     private final static Set nanoporeFolderTypes = [

src/main/groovy/life/qbic/datamodel/datasets/datastructure/files/nanopore/PoreActivityLog.groovy

@@ -0,0 +1,32 @@
+package life.qbic.datamodel.datasets.datastructure.files.nanopore
+
+import life.qbic.datamodel.datasets.datastructure.files.DataFile
+
+/**
+ * A specialisation of a DataFile, represents an Oxford Nanopore pore activity log file
+ *
+ */
+class PoreActivityLog extends DataFile {
+
+    final private static String FILE_TYPE = "csv"
+
+    final private static String NAME_SCHEMA = $/pore_activity_.*/$
+
+    protected PoreActivityLog() {}
+
+    protected PoreActivityLog(String name, String relativePath) {
+        super(name, relativePath, FILE_TYPE)
+        validateName()
+    }
+
+    static PoreActivityLog create(String name, String relativePath) {
+        return new PoreActivityLog(name, relativePath)
+    }
+
+    private void validateName() {
+        if (!(this.name =~ NAME_SCHEMA)) {
+            throw new IllegalArgumentException("Name must match the Nanopore pore activity log name schema!")
+        }
+    }
+
+}

src/main/groovy/life/qbic/datamodel/datasets/datastructure/files/nanopore/PoreScanDataLog.groovy

@@ -0,0 +1,32 @@
+package life.qbic.datamodel.datasets.datastructure.files.nanopore
+
+import life.qbic.datamodel.datasets.datastructure.files.DataFile
+
+/**
+ * A specialisation of a DataFile, represents an Oxford Nanopore pore scan data log file
+ *
+ */
+class PoreScanDataLog extends DataFile {
+
+    final private static String FILE_TYPE = "csv"
+
+    final private static String NAME_SCHEMA = $/pore_scan_data_.*/$
+
+    protected PoreScanDataLog() {}
+
+    protected PoreScanDataLog(String name, String relativePath) {
+        super(name, relativePath, FILE_TYPE)
+        validateName()
+    }
+
+    static PoreScanDataLog create(String name, String relativePath) {
+        return new PoreScanDataLog(name, relativePath)
+    }
+
+    private void validateName() {
+        if (!(this.name =~ NAME_SCHEMA)) {
+            throw new IllegalArgumentException("Name must match the Nanopore pore scan data log name schema!")
+        }
+    }
+
+}

src/main/groovy/life/qbic/datamodel/datasets/datastructure/files/nanopore/ReportHTMLLog.groovy

@@ -0,0 +1,31 @@
+package life.qbic.datamodel.datasets.datastructure.files.nanopore
+
+import life.qbic.datamodel.datasets.datastructure.files.DataFile
+
+/**
+ *  A specialisation of a DataFile, represents an Oxford Nanopore report HTML log file
+ *
+ */
+class ReportHTMLLog extends DataFile {
+
+    final private static String FILE_TYPE = "html"
+
+    final private static String NAME_SCHEMA = $/report_.*/$
+
+    protected ReportHTMLLog() {}
+
+    protected ReportHTMLLog(String name, String relativePath) {
+        super(name, relativePath, FILE_TYPE)
+        validateName()
+    }
+
+    static ReportHTMLLog create(String name, String relativePath) {
+        return new ReportHTMLLog(name, relativePath)
+    }
+
+    private void validateName() {
+        if (!(this.name =~ NAME_SCHEMA)) {
+            throw new IllegalArgumentException("Name must match the Nanopore report name schema!")
+        }
+    }
+}

src/main/groovy/life/qbic/datamodel/datasets/datastructure/files/nanopore/ReportJSONLog.groovy

@@ -0,0 +1,31 @@
+package life.qbic.datamodel.datasets.datastructure.files.nanopore
+
+import life.qbic.datamodel.datasets.datastructure.files.DataFile
+
+/**
+ *  A specialisation of a DataFile, represents an Oxford Nanopore report JSON log file
+ *
+ */
+class ReportJSONLog extends DataFile {
+
+    final private static String FILE_TYPE = "json"
+
+    final private static String NAME_SCHEMA = $/report_.*/$
+
+    protected ReportJSONLog() {}
+
+    protected ReportJSONLog(String name, String relativePath) {
+        super(name, relativePath, FILE_TYPE)
+        validateName()
+    }
+
+    static ReportJSONLog create(String name, String relativePath) {
+        return new ReportJSONLog(name, relativePath)
+    }
+
+    private void validateName() {
+        if (!(this.name =~ NAME_SCHEMA)) {
+            throw new IllegalArgumentException("Name must match the Nanopore report name schema!")
+        }
+    }
+}

src/main/groovy/life/qbic/datamodel/datasets/datastructure/files/nanopore/SampleSheetLog.groovy

@@ -0,0 +1,33 @@
+package life.qbic.datamodel.datasets.datastructure.files.nanopore
+
+import life.qbic.datamodel.datasets.datastructure.files.DataFile
+
+/**
+ *  A specialisation of a DataFile, represents an Oxford Nanopore sample sheet log file
+ *
+ * @author: Andreas Friedrich
+ */
+class SampleSheetLog extends DataFile {
+
+    final private static String FILE_TYPE = "csv"
+
+    final private static String NAME_SCHEMA = $/sample_sheet_.*/$
+
+    protected SampleSheetLog() {}
+
+    protected SampleSheetLog(String name, String relativePath) {
+        super(name, relativePath, FILE_TYPE)
+        validateName()
+    }
+
+    static SampleSheetLog create(String name, String relativePath) {
+        return new SampleSheetLog(name, relativePath)
+    }
+
+    private void validateName() {
+        if (!(this.name =~ NAME_SCHEMA)) {
+            throw new IllegalArgumentException("Name must match the Nanopore sample sheet log name schema!")
+        }
+    }
+
+}

src/main/groovy/life/qbic/datamodel/instruments/OxfordNanoporeInstrumentOutputV3.groovy

@@ -0,0 +1,22 @@
+package life.qbic.datamodel.instruments
+
+
+/**
+ * Represents the Nanopore instrument output data structure schema.
+ *
+ * The original schema is defined in as resource and is
+ * referenced here, wrapped in a Groovy class for reference
+ * in applications that want to validate the instrument
+ * output structure against the schema.
+ *
+ * @author Sven Fillinger
+ * @since 1.9.0
+ */
+class OxfordNanoporeInstrumentOutputV3 {
+
+    private static final String SCHEMA_PATH = "/schemas/nanopore-instrument-output_v3.schema.json"
+
+    static InputStream getSchemaAsStream() {
+        return OxfordNanoporeInstrumentOutputV3.getResourceAsStream(SCHEMA_PATH)
+    }
+}

src/main/java/life/qbic/datamodel/experiments/ExperimentType.java

@@ -22,5 +22,5 @@
  *
  */
 public enum ExperimentType {
-  Q_WF_NGS_QUALITYCONTROL, Q_BMI_GENERIC_IMAGING, Q_WF_MS_QUALITYCONTROL, Q__WF_NGS_MAPPING, Q_WF_MS_MAXQUANT, Q_NGS_VARIANT_CALLING, Q_NGS_MEASUREMENT, Q_NGS_MAPPING, Q_NGS_IMMUNE_MONITORING, Q_NGS_HLATYPING, Q_NGS_FLOWCELL_RUN, Q_NGS_EPITOPE_PREDICTION, Q_WF_NGS_EPITOPE_PREDICTION, Q_WF_NGS_VARIANT_ANNOTATION, Q_WF_NGS_HLATYPING, Q_EXPERIMENTAL_DESIGN, Q_SAMPLE_EXTRACTION, Q_SAMPLE_PREPARATION, Q_MHC_LIGAND_EXTRACTION, Q_MS_MEASUREMENT, Q_NGS_SINGLE_SAMPLE_RUN, Q_PROJECT_DETAILS, Q_MICROARRAY_MEASUREMENT
+  Q_WF_NGS_QUALITYCONTROL, Q_BMI_GENERIC_IMAGING, Q_WF_MS_QUALITYCONTROL, Q__WF_NGS_MAPPING, Q_WF_MS_MAXQUANT, Q_NGS_VARIANT_CALLING, Q_NGS_MEASUREMENT, Q_NGS_MAPPING, Q_NGS_IMMUNE_MONITORING, Q_NGS_HLATYPING, Q_NGS_FLOWCELL_RUN, Q_NGS_EPITOPE_PREDICTION, Q_WF_NGS_EPITOPE_PREDICTION, Q_WF_NGS_VARIANT_ANNOTATION, Q_WF_NGS_HLATYPING, Q_EXPERIMENTAL_DESIGN, Q_SAMPLE_EXTRACTION, Q_SAMPLE_PREPARATION, Q_MHC_LIGAND_EXTRACTION, Q_MS_MEASUREMENT, Q_NGS_SINGLE_SAMPLE_RUN, Q_PROJECT_DETAILS, Q_MICROARRAY_MEASUREMENT, Q_DIGIWEST_MEASUREMENT
 }

src/main/java/life/qbic/datamodel/samples/SampleType.java

@@ -22,6 +22,6 @@
  *
  */
 public enum SampleType {
-  Q_ATTACHMENT_SAMPLE, Q_BIOLOGICAL_ENTITY, Q_BIOLOGICAL_SAMPLE, Q_BMI_GENERIC_IMAGING_RUN, Q_EDDA_BENCHMARK, Q_EXT_MS_QUALITYCONTROL_RUN, Q_EXT_NGS_QUALITYCONTROL_RUN, Q_FASTA, Q_HT_QPCR_RUN, Q_MHC_LIGAND_EXTRACT, Q_MICROARRAY_RUN, Q_MS_RUN, Q_NGS_EPITOPES, Q_NGS_FLOWCELL_RUN, Q_NGS_HLATYPING, Q_NGS_IMMUNE_MONITORING, Q_NGS_IONTORRENT_RUN, Q_NGS_MAPPING, Q_NGS_MTB_DIAGNOSIS_RUN, Q_NGS_READ_MATCH_ALIGNMENT_RUN, Q_NGS_SINGLE_SAMPLE_RUN, Q_NGS_VARIANT_CALLING, Q_TEST_SAMPLE, Q_VACCINE_CONSTRUCT, Q_WF_MA_QUALITYCONTROL_RUN, Q_WF_MS_INDIVIDUALIZED_PROTEOME_RUN, Q_WF_MS_LIGANDOMICS_ID_RUN, Q_WF_MS_LIGANDOMICS_QC_RUN, Q_WF_MS_MAXQUANT_RUN, Q_WF_MS_PEPTIDEID_RUN, Q_WF_MS_QUALITYCONTROL_RUN, Q_WF_NGS_16S_TAXONOMIC_PROFILING, Q_WF_NGS_EPITOPE_PREDICTION_RUN, Q_WF_NGS_HLATYPING_RUN, Q_WF_NGS_MAPPING_RUN, Q_WF_NGS_QUALITYCONTROL_RUN, Q_WF_NGS_RNA_EXPRESSION_ANALYSIS_RUN, Q_WF_NGS_SHRNA_COUNTING_RUN, Q_WF_NGS_VARIANT_ANNOTATION_RUN, Q_WF_NGS_VARIANT_CALLING_RUN
+  Q_ATTACHMENT_SAMPLE, Q_BIOLOGICAL_ENTITY, Q_BIOLOGICAL_SAMPLE, Q_BMI_GENERIC_IMAGING_RUN, Q_EDDA_BENCHMARK, Q_EXT_MS_QUALITYCONTROL_RUN, Q_EXT_NGS_QUALITYCONTROL_RUN, Q_FASTA, Q_HT_QPCR_RUN, Q_MHC_LIGAND_EXTRACT, Q_MICROARRAY_RUN, Q_MS_RUN, Q_NGS_EPITOPES, Q_NGS_FLOWCELL_RUN, Q_NGS_HLATYPING, Q_NGS_IMMUNE_MONITORING, Q_NGS_IONTORRENT_RUN, Q_NGS_MAPPING, Q_NGS_MTB_DIAGNOSIS_RUN, Q_NGS_READ_MATCH_ALIGNMENT_RUN, Q_NGS_SINGLE_SAMPLE_RUN, Q_NGS_VARIANT_CALLING, Q_TEST_SAMPLE, Q_VACCINE_CONSTRUCT, Q_WF_MA_QUALITYCONTROL_RUN, Q_WF_MS_INDIVIDUALIZED_PROTEOME_RUN, Q_WF_MS_LIGANDOMICS_ID_RUN, Q_WF_MS_LIGANDOMICS_QC_RUN, Q_WF_MS_MAXQUANT_RUN, Q_WF_MS_PEPTIDEID_RUN, Q_WF_MS_QUALITYCONTROL_RUN, Q_WF_NGS_16S_TAXONOMIC_PROFILING, Q_WF_NGS_EPITOPE_PREDICTION_RUN, Q_WF_NGS_HLATYPING_RUN, Q_WF_NGS_MAPPING_RUN, Q_WF_NGS_QUALITYCONTROL_RUN, Q_WF_NGS_RNA_EXPRESSION_ANALYSIS_RUN, Q_WF_NGS_SHRNA_COUNTING_RUN, Q_WF_NGS_VARIANT_ANNOTATION_RUN, Q_WF_NGS_VARIANT_CALLING_RUN, Q_DIGIWEST_RUN
 
 }

src/main/resources/schemas/nanopore-instrument-output.schema.json

@@ -402,6 +402,9 @@
                   {
                     "$ref": "#/definitions/report_pdf_log"
                   },
+                  {
+                    "$ref": "#/definitions/report_html_log"
+                  },
                   {
                     "$ref": "#/definitions/sequencing_summary_log"
                   },
@@ -517,6 +520,23 @@
         }
       ]
     },
+    "report_html_log": {
+      "allOf": [
+        {
+          "$ref": "#/definitions/file"
+        },
+        {
+          "properties": {
+            "name": {
+              "pattern": "report_.*"
+            },
+            "file_type": {
+              "pattern": "html"
+            }
+          }
+        }
+      ]
+    },
     "sequencing_summary_log": {
       "allOf": [
         {