add update

tcr_pmhc_complexes.py

@@ -95,6 +95,50 @@ class FastaHeaderInfo(BaseModel):
 
         return cls(pdb_id=pdb_id, chain_ids=chain_ids, auth_chain_ids=auth_chain_ids, description=description, is_polymeric=is_polymeric)
 
+# 对于标准fasta的兼容尝试处理，废弃
+# class FastaHeaderInfo(BaseModel):
+#     info: Dict[str, Union[str, List[str], Dict[str, str]]] = {}
+#     is_polymeric: Optional[str] = None  # 新增属性
+
+#     @classmethod
+#     def from_header_line(cls, header_line: str):
+#         header_info = {}
+#         if '|' in header_line and 'Chain' in header_line:
+#             # PDB FASTA格式处理
+#             parts = header_line.split('|')
+#             header_info['identifier'] = parts[0].strip('>')
+#             chain_info = parts[1] if len(parts) > 1 else ''
+#             header_info['description'] = parts[2] if len(parts) > 2 else ''
+
+#             chain_parts = chain_info.replace('Chain ', '').replace('Chains ', '').split(',')
+#             chain_ids = []
+#             auth_chain_ids = {}
+
+#             for part in chain_parts:
+#                 part = part.strip()
+#                 if '[' in part:
+#                     chain_id, auth_chain_id = part.split('[')
+#                     chain_id = chain_id.strip()
+#                     auth_chain_id = auth_chain_id.strip(']').strip()
+#                     chain_ids.append(chain_id)
+#                     auth_chain_ids[chain_id] = auth_chain_id
+#                 else:
+#                     chain_ids.append(part)
+#                     auth_chain_ids[part] = part
+
+#             header_info['chain_ids'] = chain_ids
+#             header_info['auth_chain_ids'] = auth_chain_ids
+#             header_info['is_polymeric'] = "Yes" if len(chain_ids) > 1 else "No"
+#         else:
+#             # 处理非PDB或非标准FASTA头信息
+#             identifier = header_line[1:].split()[0]  # 取第一个空格前的文本作为标识符
+#             description = ' '.join(header_line[1:].split()[1:])  # 剩余的文本作为描述
+#             header_info['identifier'] = identifier
+#             header_info['description'] = description
+#             header_info['is_polymeric'] = "Unknown"
+
+#         return cls(info=header_info)
+
 
 class FastaSequence(BaseModel):
     header_info: FastaHeaderInfo