bugfixes on analysis-run dumper

EBI-Metagenomics · Jan 5, 2024 · f36021e · f36021e
1 parent d185d4e
commit f36021e
Show file tree

Hide file tree

Showing 4 changed files with 24 additions and 13 deletions.
diff --git a/emgapi/management/commands/ebi_search_analysis_dump.py b/emgapi/management/commands/ebi_search_analysis_dump.py
@@ -59,15 +59,15 @@ def get_analysis_context(self, analysis: AnalysisJob):
 
         try:
             go_annotation: Optional[AnalysisJobGoTerm] = AnalysisJobGoTerm.objects.get(
-                pk=analysis.job_id
+                pk=str(analysis.job_id)
             )
         except AnalysisJobGoTerm.DoesNotExist:
             logger.debug(f"Could not find go terms for {analysis.job_id}")
             go_annotation = None
 
         try:
             ips_annotation: Optional[AnalysisJobInterproIdentifier] = AnalysisJobInterproIdentifier.objects.get(
-                pk=analysis.job_id
+                pk=str(analysis.job_id)
             )
         except AnalysisJobInterproIdentifier.DoesNotExist:
             logger.debug(f"Could not find IPS terms for {analysis.job_id}")
@@ -203,10 +203,10 @@ def handle(self, *args, **options):
                     )
                 )
 
-        nowish = timezone.now() + timedelta(minutes=1)
-        # Small buffer into the future so that the indexing time remains ahead of auto-now updated times.
+            nowish = timezone.now() + timedelta(minutes=1)
+            # Small buffer into the future so that the indexing time remains ahead of auto-now updated times.
 
-        for analysis in analyses:
-            analysis.last_indexed = nowish
+            for analysis in page:
+                analysis.last_indexed = nowish
 
-        AnalysisJob.objects.bulk_update(analyses, fields=["last_indexed"])
+            AnalysisJob.objects.bulk_update(page, fields=["last_indexed"])
diff --git a/emgapi/templates/ebi_search/analysis.xml b/emgapi/templates/ebi_search/analysis.xml
@@ -7,9 +7,10 @@
                 <date type="sample_collection_date" value="{{ analysis.sample.collection_date|date:'Y-m-d' }}"/>
             </dates>
             <additional_fields>
-                <field name="experiment_type">assembly</field>
+                <field name="experiment_type">{{ analysis.experiment_type.experiment_type }}</field>
                 <field name="pipeline_version">{{ analysis.pipeline.release_version }}</field>
                 <field name="sample_name">{{ analysis.sample.sample_name | escape }}</field>
+                <field name="sample_description">{{ analysis.sample.sample_desc | escape }}</field>
                 <field name="project_name">{{ analysis.study.study_name | escape }}</field>
                 <field name="biome_name">{{ analysis.study.biome.biome_name | escape }}</field>
 
@@ -66,16 +67,26 @@
                 <ref dbkey="{{ analysis.study.accession }}" dbname="metagenomics_projects"/>
                 <ref dbkey="{{ analysis.study.project_id }}" dbname="ena_project"/>
 
+                {# biosample #}
+                {% if analysis.sample.primary_accession %}
+                <ref dbkey="{{ analysis.sample.primary_accession }}" dbname="biosamples"/>
+                {% endif %}
+
+                {# INSDC sample like DRSxx, ERSxx, SRSxx #}
+                {% if analysis.sample.accession and analysis.sample.accession|slice:"1:3" == "RS"%}
+                <ref dbkey="{{ analysis.sample.accession }}" dbname="sra-sample"/>
+                {% endif %}
+
                 {# Assembly analysis #}
-                {% if analysis.assembly and analysis.assembly.accession|slice:":4" == "GCA_" %}
-                <ref dbkey="{{ analysis.assembly.legacy_accession }}" dbname="ena_wgs_sequence_set"/>
-                <ref dbkey="{{ analysis.assembly.accession }}" dbname="assembly"/>
+                {% if analysis.assembly %}
+                <ref dbkey="{{ analysis.assembly.accession }}" dbname="analysis"/>
                 {% endif %}
 
                 {# WGS/Amplicon analysis #}
                 {% if analysis.run %}
                 <ref dbkey="{{ analysis.run.accession }}" dbname="ena_run"/>
                 {% endif %}
+
                 {% for go in analysis_go_entries %}
                 <ref dbkey="{{ go.accession }}" dbname="go"/>
                 {% endfor %}

diff --git a/emgcli/__init__.py b/emgcli/__init__.py
@@ -1 +1 @@
-__version__: str = "2.4.40"
+__version__: str = "2.4.41"
diff --git a/pyproject.toml b/pyproject.toml
@@ -117,7 +117,7 @@ max-line-length = 119
 """
 
 [tool.bumpversion]
-current_version = "2.4.40"
+current_version = "2.4.41"
 
 [[tool.bumpversion.files]]
 filename = "emgcli/__init__.py"