Upload app.py with huggingface_hub

app.py

@@ -200,11 +200,14 @@ def get_confident_alleles(bead_results, bead_hla_map):
     例: A*11:01 在 X2 → 排除 A*11:01
         A*11:02 只在 X8 → 保留 A*11:02 → 報告寫 A11(A*11:02)
     """
+    def _clean(ag):
+        return ag.split('/')[0] if '/' in ag else ag
+
     x6x8 = set()
     not_x6x8 = set()  # Rxn=1, 2, 4 全部排除
     for bid, r in bead_results.items():
         hla = bead_hla_map.get(bid, {})
-        alleles = _parse_allele_list(hla.get('allele', ''))
+        alleles = {_clean(a) for a in _parse_allele_list(hla.get('allele', ''))}
         if r['rxn'] >= 6:
             x6x8.update(alleles)
         else:
@@ -259,7 +262,11 @@ def _build_allele_to_sero(bead_hla_map):
             alleles = allele_by_locus.get(locus, [])
             for i, ag in enumerate(alleles):
                 if i < len(seros):
+                    # 同時建立原始和清理後的 mapping
                     mapping[ag] = seros[i]
+                    clean_ag = ag.split('/')[0] if '/' in ag else ag
+                    if clean_ag != ag:
+                        mapping[clean_ag] = seros[i]
     return mapping
 
 
@@ -366,19 +373,30 @@ def generate_specificity(confident_alleles, bead_hla_map):
             if sero:
                 all_alleles_per_sero.setdefault(sero, set()).add(ag)
 
-    # 分組: sero groups + DQA1/DQB1/DPA1/DPB1 獨立
+    # 清理 allele 名稱（移除 /276N, /163N 等後綴）
+    def clean_allele(ag):
+        return ag.split('/')[0] if '/' in ag else ag
+
+    # 分組:
+    # DQA1/DPA1 (alpha chain, 無 sero) → allele 格式
+    # DQB1/DPB1 (beta chain, 有 sero) → 回歸 sero 對照
+    # 其他 → sero 對照
     sero_groups = {}
-    allele_groups = {'DQA1': set(), 'DQB1': set(), 'DPA1': set(), 'DPB1': set()}
+    allele_groups = {'DQA1': set(), 'DPA1': set()}
 
     for ag in confident_alleles:
-        # DQA1/DQB1/DPA1/DPB1 alleles 獨立分組
-        prefix = ag.split('*')[0] if '*' in ag else ''
+        ag_clean = clean_allele(ag)
+        prefix = ag_clean.split('*')[0] if '*' in ag_clean else ''
         if prefix in allele_groups:
-            allele_groups[prefix].add(ag)
+            # DQA1/DPA1 → allele 格式
+            allele_groups[prefix].add(ag_clean)
         else:
+            # DQB1/DPB1 和其他 → sero 對照
             sero = a2s.get(ag)
+            if not sero:
+                sero = a2s.get(ag_clean)
             if sero:
-                sero_groups.setdefault(sero, set()).add(ag)
+                sero_groups.setdefault(sero, set()).add(ag_clean)
 
     # 排序: 按 locus 再按數字
     import re as _re
@@ -420,12 +438,17 @@ def generate_specificity(confident_alleles, bead_hla_map):
             # 兩個以上 allele → 全部列出
             parts.append(f'{sero}({" ".join(sorted(conf))})')
 
-    # DQA1/DQB1/DPA1/DPB1 allele groups
-    for prefix in ['DQA1', 'DQB1', 'DPA1', 'DPB1']:
+    # DQA1/DPA1 allele groups (alpha chain, 無 sero 對照)
+    for prefix in ['DQA1', 'DPA1']:
         alleles = allele_groups[prefix]
         if alleles:
-            short = sorted('*' + a.split('*')[1] for a in alleles)
-            parts.append(f'{prefix}({" ".join(short)})')
+            if len(alleles) == 1:
+                # 單一 allele → 不加括號: DQA1*01:03
+                parts.append(list(alleles)[0])
+            else:
+                # 多個 → 加括號: DQA1(*01:03 *05:05)
+                short = sorted('*' + a.split('*')[1] for a in alleles)
+                parts.append(f'{prefix}({" ".join(short)})')
 
     return ' '.join(parts)