Add latest changes

ave-dcd · Dec 5, 2024 · fa54de7 · fa54de7
1 parent af28a0f
commit fa54de7
Show file tree

Hide file tree

Showing 5 changed files with 40 additions and 40 deletions.
diff --git a/schema.json b/schema.json
@@ -370,11 +370,7 @@
                     "title": "Version"
                 },
                 "code": {
-                    "allOf": [
-                        {
-                            "$ref": "#/$defs/Code"
-                        }
-                    ],
+                    "$ref": "#/$defs/Code",
                     "description": "A symbol uniquely identifying the concept, as in a syntax defined by the code system. CURIE format is preferred where possible (e.g. 'SO:0000704' is the CURIE form of the Sequence Ontology code for 'gene')."
                 }
             },
@@ -412,19 +408,11 @@
             "description": "A mapping to a concept in a terminology or code system.",
             "properties": {
                 "coding": {
-                    "allOf": [
-                        {
-                            "$ref": "#/$defs/Coding"
-                        }
-                    ],
+                    "$ref": "#/$defs/Coding",
                     "description": "A structured representation of a code for a defined concept in a terminology or code system."
                 },
                 "relation": {
-                    "allOf": [
-                        {
-                            "$ref": "#/$defs/Relation"
-                        }
-                    ],
+                    "$ref": "#/$defs/Relation",
                     "description": "A mapping relation between concepts as defined by the Simple Knowledge Organization System (SKOS)."
                 }
             },
@@ -439,11 +427,7 @@
             "description": "Representation of a variation by a specified nomenclature or syntax for a\nVariation object. Common examples of expressions for the description of molecular\nvariation include the HGVS and ISCN nomenclatures.",
             "properties": {
                 "syntax": {
-                    "allOf": [
-                        {
-                            "$ref": "#/$defs/Syntax"
-                        }
-                    ],
+                    "$ref": "#/$defs/Syntax",
                     "description": "The syntax used to describe the variation. The value should be one of the supported syntaxes."
                 },
                 "value": {
@@ -751,11 +735,7 @@
                     "title": "Mappings"
                 },
                 "sequence": {
-                    "allOf": [
-                        {
-                            "$ref": "#/$defs/SequenceString"
-                        }
-                    ],
+                    "$ref": "#/$defs/SequenceString",
                     "description": "the literal sequence"
                 }
             },
@@ -972,21 +952,24 @@
                 "pre_mapped": {
                     "anyOf": [
                         {
-                            "$ref": "#/$defs/Allele"
+                            "$ref": "#/$defs/CisPhasedBlock"
                         },
                         {
-                            "$ref": "#/$defs/CisPhasedBlock"
+                            "$ref": "#/$defs/Allele"
                         }
                     ],
                     "title": "Pre Mapped"
                 },
                 "post_mapped": {
                     "anyOf": [
+                        {
+                            "$ref": "#/$defs/CisPhasedBlock"
+                        },
                         {
                             "$ref": "#/$defs/Allele"
                         },
                         {
-                            "$ref": "#/$defs/CisPhasedBlock"
+                            "type": "null"
                         }
                     ],
                     "title": "Post Mapped"

diff --git a/src/dcd_mapping/align.py b/src/dcd_mapping/align.py
@@ -165,9 +165,16 @@ def _get_blat_output(metadata: ScoresetMetadata, silent: bool) -> QueryResult:
     """
     with tempfile.NamedTemporaryFile() as query_file:
         query_file = _build_query_file(metadata, Path(query_file.name))
+        if len(metadata.target_sequence) > 25000:
+            msg = f"Target sequence for {metadata.urn} must have a length <= 25000 to run BLAT"
+            raise AlignmentError(msg)
+
         if metadata.target_sequence_type == TargetSequenceType.PROTEIN:
             target_args = "-q=prot -t=dnax"
-        elif metadata.target_gene_category == TargetType.PROTEIN_CODING:
+        elif (
+            metadata.target_gene_category == TargetType.PROTEIN_CODING
+            and len(metadata.target_sequence) <= 10000
+        ):
             target_args = "-q=dnax -t=dnax"
         else:
             target_args = ""

diff --git a/src/dcd_mapping/annotate.py b/src/dcd_mapping/annotate.py
@@ -21,6 +21,7 @@
     CisPhasedBlock,
     Expression,
     LiteralSequenceExpression,
+    SequenceString,
 )
 
 from dcd_mapping.lookup import (
@@ -115,7 +116,7 @@ def _get_vrs_ref_allele_seq(
         ref = sr.get_sequence(seq, start, end)
         if ref is None:
             raise ValueError
-    return Extension(name="vrs_ref_allele_seq", value=ref)
+    return SequenceString(root=ref)
 
 
 def _get_hgvs_string(allele: Allele, accession: str) -> tuple[str, Syntax]:
@@ -208,9 +209,11 @@ def _annotate_allele_mapping(
 
     # get vrs_ref_allele_seq for pre-mapped variants
     pre_mapped.extensions = [
-        _get_vrs_ref_allele_seq(pre_mapped, metadata, tx_results),
         _get_vrs_1_3_ext(pre_mapped),
     ]
+    pre_mapped.location.sequence = _get_vrs_ref_allele_seq(
+        pre_mapped, metadata, tx_results
+    )
 
     # Determine reference sequence
     if mapped_score.annotation_layer == AnnotationLayer.GENOMIC:
@@ -228,9 +231,10 @@ def _annotate_allele_mapping(
     sr = get_seqrepo()
     loc = mapped_score.post_mapped.location
     sequence_id = f"ga4gh:{loc.sequenceReference.refgetAccession}"
-    ref = sr.get_sequence(sequence_id, loc.start, loc.end)
+    post_mapped.location.sequence = SequenceString(
+        root=sr.get_sequence(sequence_id, loc.start, loc.end)
+    )
     post_mapped.extensions = [
-        Extension(name="vrs_ref_allele_seq", value=ref),
         _get_vrs_1_3_ext(post_mapped),
         _get_va_digest(pre_mapped),
     ]
@@ -246,6 +250,10 @@ def _annotate_allele_mapping(
             post_mapped if _is_valid_allele(pre_mapped, align_result) else None
         )
 
+    # Remove extra digest attributes
+    pre_mapped.digest = None
+    post_mapped.digest = None
+
     return ScoreAnnotationWithLayer(
         pre_mapped=pre_mapped,
         post_mapped=post_mapped,
@@ -280,9 +288,10 @@ def _annotate_cpb_mapping(
     # get vrs_ref_allele_seq for pre-mapped variants
     for allele in pre_mapped.members:
         allele.extensions = [
-            _get_vrs_ref_allele_seq(allele, metadata, tx_results),
             _get_vrs_1_3_ext(allele),
         ]
+        allele.location.sequence = _get_vrs_ref_allele_seq(allele, metadata, tx_results)
+        allele.digest = None
     # Determine reference sequence
     if mapping.annotation_layer == AnnotationLayer.GENOMIC:
         sequence_id = (
@@ -305,9 +314,10 @@ def _annotate_cpb_mapping(
     ):
         loc = post_mapped_allele.location
         sequence_id = f"ga4gh:{loc.sequenceReference.refgetAccession}"
-        ref = sr.get_sequence(sequence_id, loc.start, loc.end)
+        post_mapped_allele.location.sequence = SequenceString(
+            root=sr.get_sequence(sequence_id, loc.start, loc.end)
+        )
         post_mapped_allele.extensions = [
-            Extension(name="vrs_ref_allele_seq", value=ref),
             _get_vrs_1_3_ext(post_mapped_allele),
             _get_va_digest(pre_mapped_allele),
         ]
@@ -317,6 +327,7 @@ def _annotate_cpb_mapping(
             pre_mapped_allele, align_result
         ):
             valid_post_mapped_alleles.append(post_mapped_allele)
+        post_mapped_allele.digest = None
     post_mapped.members = valid_post_mapped_alleles
 
     pre_mapped.extensions = [

diff --git a/src/dcd_mapping/schemas.py b/src/dcd_mapping/schemas.py
@@ -18,9 +18,9 @@ class TargetSequenceType(str, Enum):
 class TargetType(str, Enum):
     """Define target gene types."""
 
-    PROTEIN_CODING = "Protein coding"
-    REGULATORY = "Regulatory"
-    OTHER_NC = "Other noncoding"
+    PROTEIN_CODING = "protein_coding"
+    REGULATORY = "regulatory"
+    OTHER_NC = "other_noncoding"
 
 
 class UniProtRef(BaseModel):

diff --git a/src/dcd_mapping/vrs_map.py b/src/dcd_mapping/vrs_map.py
@@ -462,7 +462,6 @@ def _get_variation(
 
         # Run ga4gh_identify to assign VA digest
         allele.id = ga4gh_identify(allele)
-        allele.digest = None
         alleles.append(allele)
 
     if not alleles: