IGVF-2113-index-file

src/igvfd/audit/file.py

@@ -115,47 +115,3 @@ def audit_external_reference_files(value, system):
                 f'but does not have identifier(s) from an external resource listed in `dbxrefs`.'
             )
             yield AuditFailure(audit_message.get('audit_category', ''), f'{detail} {audit_message.get("audit_description", "")}', level=audit_message.get('audit_level', ''))
-
-
-@audit_checker('AlignmentFile', frame='object')
-def audit_bai_alignment_files(value, system):
-    '''
-    [
-        {
-            "audit_description": "Alignment files in bai format are expected to have their corresponding bam file in `derived_from`.",
-            "audit_category": "incorrect bam file",
-            "audit_level": "ERROR"
-        }
-    ]
-    '''
-    audit_message = get_audit_message(audit_bai_alignment_files)
-    object_type = space_in_words(value['@type'][0]).capitalize()
-    check_properties_list = ['content_type', 'assembly', 'filtered', 'redacted', 'transcriptome_annotation']
-    inconsistent_properties_list = []
-    if value.get('file_format') == 'bai':
-        if 'derived_from' not in value:
-            detail = (
-                f'{object_type} {audit_link(path_to_text(value["@id"]), value["@id"])} has no bam file in `derived_from`.')
-            yield AuditFailure(audit_message.get('audit_category', ''), f'{detail} {audit_message.get("audit_description", "")}', level=audit_message.get('audit_level', ''))
-        else:
-            derived_from_file = value.get('derived_from')
-            if len(derived_from_file) > 1:
-                detail = (
-                    f'{object_type} {audit_link(path_to_text(value["@id"]), value["@id"])} has multiple files in `derived_from`.')
-                yield AuditFailure(audit_message.get('audit_category', ''), f'{detail} {audit_message.get("audit_description", "")}', level=audit_message.get('audit_level', ''))
-            else:
-                derived_from_file_obj = system.get('request').embed(
-                    derived_from_file[0], '@@object?skip_calculated=true')
-                if derived_from_file_obj.get('file_format') != 'bam':
-                    detail = (
-                        f'{object_type} {audit_link(path_to_text(value["@id"]), value["@id"])} has incorrect file in `derived_from`.')
-                    yield AuditFailure(audit_message.get('audit_category', ''), f'{detail} {audit_message.get("audit_description", "")}', level=audit_message.get('audit_level', ''))
-                else:
-                    for property in check_properties_list:
-                        if value.get(property) != derived_from_file_obj.get(property):
-                            inconsistent_properties_list.append(property)
-                    if inconsistent_properties_list:
-                        inconsistent_properties_str = ', '.join(inconsistent_properties_list)
-                        detail = (f'{object_type} {audit_link(path_to_text(value["@id"]), value["@id"])} has the following inconsistent properties with its bam file in `derived_from`: '
-                                  f'{inconsistent_properties_str}.')
-                        yield AuditFailure(audit_message.get('audit_category', ''), f'{detail} {audit_message.get("audit_description", "")}', level=audit_message.get('audit_level', ''))

src/igvfd/audit/index_file.py

@@ -0,0 +1,39 @@
+from snovault.auditor import (
+    audit_checker,
+    AuditFailure,
+)
+from .formatter import (
+    audit_link,
+    path_to_text,
+    get_audit_message,
+    space_in_words
+)
+
+
+@audit_checker('IndexFile', frame='object')
+def audit_index_files_derived_from(value, system):
+    '''
+    [
+        {
+            "audit_description": "Index files in tbi format are expected to have a corresponding tsv or vcf file in `derived_from`.",
+            "audit_category": "unexpected indexed file",
+            "audit_level": "ERROR"
+        }
+    ]
+    '''
+    audit_message_tbi = get_audit_message(audit_index_files_derived_from)
+    object_type = space_in_words(value['@type'][0]).capitalize()
+    # For tbi files, check that the indexed file is of an expected file_format.
+    # No need to check bai files, since Alignment Files can only be bams.
+    derived_from_file = value.get('derived_from', [])
+    derived_from_file_obj = system.get('request').embed(derived_from_file[0], '@@object?skip_calculated=true')
+    if value['file_format'] == 'tbi':
+        if derived_from_file_obj.get('file_format') not in ['tsv', 'vcf']:
+            detail = (
+                f'{object_type} {audit_link(path_to_text(value["@id"]), value["@id"])} '
+                f'has a file of unexpected file format in `derived_from`.')
+            yield AuditFailure(
+                audit_message_tbi.get('audit_category', ''),
+                f'{detail} {audit_message_tbi.get("audit_description", "")}',
+                level=audit_message_tbi.get('audit_level', '')
+            )

src/igvfd/commands/make_audit_docstring_json.py

@@ -10,6 +10,7 @@
 import igvfd.audit.formatter
 import igvfd.audit.human_donor
 import igvfd.audit.in_vitro_system
+import igvfd.audit.index_file
 import igvfd.audit.matrix_file
 import igvfd.audit.measurement_set
 import igvfd.audit.multiplexed_sample
@@ -37,6 +38,7 @@
     igvfd.audit.formatter,
     igvfd.audit.human_donor,
     igvfd.audit.in_vitro_system,
+    igvfd.audit.index_file,
     igvfd.audit.matrix_file,
     igvfd.audit.measurement_set,
     igvfd.audit.multiplexed_sample,

src/igvfd/loadxl.py

@@ -62,6 +62,7 @@
     'matrix_file',
     'model_file',
     'tabular_file',
+    'index_file',
     'genome_browser_annotation_file',
     'image_file'
 ]

src/igvfd/mappings/alignment_file.json

@@ -1,6 +1,6 @@
 {
-    "hash": "d2d11b08bd6e06c2ba47015b17b1f8aa",
-    "index_name": "alignment_file_d2d11b08",
+    "hash": "12428c46f270cbe2fdf3b7b7867a2c68",
+    "index_name": "alignment_file_12428c46",
     "item_type": "alignment_file",
     "mapping": {
         "dynamic_templates": [