Added kaggle workflow

Author: hokinus

tools/fasta/chain_parser.py

@@ -0,0 +1,73 @@
+import re
+from typing import Dict, List, Tuple
+
+
+def parse_fasta(fasta_content: str) -> Dict[str, Tuple[str, List[str]]]:
+    """
+    Parse FASTA content into dictionary.
+
+    Args:
+        fasta_content: Multi-line FASTA string with format:
+        >1A1T_1|Chain A[auth B]|SL3 STEM-LOOP RNA|
+        or
+        >104D_1|Chains A[auth A], B[auth B]|DNA/RNA (...)|
+
+    Returns:
+        Dictionary mapping auth chain_id to (sequence, list_of_auth_chain_ids)
+        Example: {"A": ("ACGT", ["A", "B"]), "C": ("UGCA", ["C"])}
+        The key is the auth chain ID, and the list contains all auth chain IDs for this sequence
+    """
+    result = {}
+    lines = fasta_content.strip().split("\n")
+
+    i = 0
+    while i < len(lines):
+        line = lines[i].strip()
+
+        if line.startswith(">"):
+            # Parse new format header: >104D_1|Chains A[auth A], B[auth B]|...| or >1A1T_1|Chain A[auth B]|...|
+            # Extract the chains part (between first | and second |)
+            parts = line.split("|")
+            if len(parts) < 2:
+                print("Warning: Malformed FASTA header:", line)
+                auth_chain_ids = []
+                chains_part = ""
+            else:
+                chains_part = parts[1].strip()
+
+                # Extract auth chain IDs from patterns like "Chain A[auth B]" or "Chains A[auth A], B[auth B] or just "Chain A" or "Chains A, B"
+                auth_chain_ids = []
+                replaced_chains_part = re.sub(r"^Chains? ", "", chains_part)
+                chains = replaced_chains_part.split(",")
+                for chain in chains:
+                    auth_match = re.search(r"\[auth ([^\]]+)\]", chain)
+                    if auth_match:
+                        auth_chain_ids.append(auth_match.group(1).strip())
+                    else:
+                        c = chain.strip()
+                        if c:
+                            auth_chain_ids.append(c)
+
+            # # Find all Y[auth X] or just Y patterns
+            # auth_matches = re.findall(r"[^ ,]+(?:\[auth ([^\]]+)\])?", chains_part)
+
+            # auth_chain_ids = [match.strip() for match in auth_matches]
+
+            if not auth_chain_ids:
+                print("Warning: Empty chains part:", chains_part)
+                primary_auth_chain = None
+            else:
+                # Use the first auth chain ID as the key
+                primary_auth_chain = auth_chain_ids[0]
+
+            # Read sequence (next lines until next header or end)
+            sequence = ""
+            while (i + 1) < len(lines) and lines[i + 1].startswith(">") is False:
+                sequence += lines[i + 1].strip()
+                i += 1
+            for chain in auth_chain_ids:
+                result[chain] = (sequence, auth_chain_ids)
+
+        i += 1
+
+    return result

validation/validate_bioassembly_metadata.py

@@ -24,7 +24,7 @@
 except ImportError:
     has_biotite = False
 # Import parse_fasta from the tools/fasta directory
-sys.path.insert(0, str(Path(__file__).parent / "tools/fasta"))
+sys.path.insert(0, str(Path(__file__).parents[1] / "tools/fasta"))
 from chain_parser import parse_fasta
 
 csv.field_size_limit(sys.maxsize)