feat: blast by cdr3

backend/antigenapi/bioinformatics/blast.py

@@ -11,14 +11,20 @@
 BLAST_NUM_THREADS = 4
 
 
-def get_db_fasta(include_run: Optional[int] = None, exclude_run: Optional[int] = None):
+def get_db_fasta(
+    include_run: Optional[int] = None,
+    exclude_run: Optional[int] = None,
+    query_type: str = "full",
+):
     """Get the sequencing database in fasta format.
 
     Args:
         include_run (int, optional): Sequencing run ID to include.
           Defaults to None.
         exclude_run (int, optional): Sequencing run ID to exclude.
           Defaults to None.
+        query_type (str): Query type - "full" sequence or "cdr3"
+          Defaults to "full".
 
     Returns:
         str: Sequencing run as a FASTA format string
@@ -31,20 +37,35 @@ def get_db_fasta(include_run: Optional[int] = None, exclude_run: Optional[int] =
         query = query.exclude(sequencing_run_id=exclude_run)
     for sr in query:
         airr_file = read_airr_file(
-            sr.airr_file, usecols=("sequence_id", "sequence_alignment_aa")
+            sr.airr_file,
+            usecols=(
+                "sequence_id",
+                "cdr3_aa" if query_type == "cdr3" else "sequence_alignment_aa",
+            ),
         )
-        airr_file = airr_file[airr_file.sequence_alignment_aa.notna()]
+        airr_file = airr_file[
+            (
+                airr_file.cdr3_aa.notna()
+                if query_type == "cdr3"
+                else airr_file.sequence_alignment_aa.notna()
+            )
+        ]
         if not airr_file.empty:
-            for _, row in airr_file.iterrows():
-                seq = row.sequence_alignment_aa.replace(".", "")
-                try:
-                    if fasta_data[row.sequence_id] != seq:
-                        raise ValueError(
-                            f"Different sequences with same name! {row.sequence_id}"
-                        )
-                    continue
-                except KeyError:
-                    fasta_data[row.sequence_id] = seq
+            if query_type == "cdr3":
+                cdr3s = set(airr_file.cdr3_aa.unique())
+                for cdr3 in cdr3s:
+                    fasta_data[f"CDR3: {cdr3}"] = cdr3
+            else:
+                for _, row in airr_file.iterrows():
+                    seq = row.sequence_alignment_aa.replace(".", "")
+                    try:
+                        if fasta_data[row.sequence_id] != seq:
+                            raise ValueError(
+                                f"Different sequences with same name! {row.sequence_id}"
+                            )
+                        continue
+                    except KeyError:
+                        fasta_data[row.sequence_id] = seq
 
     fasta_files = as_fasta_files(fasta_data, max_file_size=None)
     if fasta_files:
@@ -53,33 +74,37 @@ def get_db_fasta(include_run: Optional[int] = None, exclude_run: Optional[int] =
     return ""
 
 
-def get_sequencing_run_fasta(sequencing_run_id: int):
+def get_sequencing_run_fasta(sequencing_run_id: int, query_type: str):
     """Get sequencing run in BLAST format.
 
     Args:
         sequencing_run_id (int): Sequencing run ID
+        query_type (str): Query type - "full" sequence or "cdr3"
 
     Returns:
         str: Sequencing run as a FASTA format string
     """
-    return get_db_fasta(include_run=sequencing_run_id)
+    return get_db_fasta(include_run=sequencing_run_id, query_type=query_type)
 
 
 def run_blastp(
-    sequencing_run_id: int, outfmt: str = BLAST_FMT_MULTIPLE_FILE_BLAST_JSON
+    sequencing_run_id: int,
+    query_type: str = "full",
+    outfmt: str = BLAST_FMT_MULTIPLE_FILE_BLAST_JSON,
 ):
     """Run blastp for a sequencing run vs rest of database.
 
     Args:
         sequencing_run_id (int): Sequencing run ID.
+        query_type (str): Query type - "full" sequence or "cdr3"
 
     Returns:
         JSONResponse: Single file BLAST JSON
     """
     db_data = get_db_fasta()
     if not db_data:
         return None
-    query_data = get_sequencing_run_fasta(sequencing_run_id)
+    query_data = get_sequencing_run_fasta(sequencing_run_id, query_type=query_type)
     if not query_data:
         return None
 

backend/antigenapi/views_old.py

@@ -1062,7 +1062,10 @@ def search_sequencing_run_results(self, request, query):
     )
     def get_blast_sequencing_run(self, request, pk):
         """BLAST sequencing run vs database."""
-        blast_str = run_blastp(pk)
+        query_type = self.request.query_params.get("queryType", "full")
+        if query_type not in ("full", "cdr3"):
+            raise ValueError(f"Unknown queryType: {query_type}")
+        blast_str = run_blastp(pk, query_type=query_type)
         if not blast_str:
             return JsonResponse({"hits": []}, status=status.HTTP_404_NOT_FOUND)
 
@@ -1083,9 +1086,13 @@ def get_blast_sequencing_run(self, request, pk):
             for blast_hit_set in run_res["hits"]:
                 subject_title = blast_hit_set["description"][0]["title"]
                 query_title = run_res["query_title"]
-                query_cdr3 = airr_df.at[query_title, "cdr3_aa"]
-                if pd.isna(query_cdr3):
-                    query_cdr3 = None
+                if query_type == "cdr3":
+                    # TODO: Make more robust
+                    query_cdr3 = query_title[6:]
+                else:
+                    query_cdr3 = airr_df.at[query_title, "cdr3_aa"]
+                    if pd.isna(query_cdr3):
+                        query_cdr3 = None
                 if subject_title.strip() == query_title.strip():
                     continue
                 hsps = blast_hit_set["hsps"][0]

frontend/package.json

@@ -3,7 +3,7 @@
   "version": "0.1.0",
   "private": true,
   "dependencies": {
-    "@headlessui/react": "^1.7.19",
+    "@headlessui/react": "2.1",
     "@heroicons/react": "^2.1.5",
     "@sentry/react": "^7.119.1",
     "@sentry/tracing": "^7.114.0",