Upload bin/filter_civic_vcf.py

bin/filter_civic_vcf.py

@@ -0,0 +1,198 @@
+#!/usr/bin/env python3
+"""
+过滤CIViC VCF文件，保留核心治疗相关字段和辅助参考字段
+
+保留的字段（基于CSQ索引）：
+核心治疗相关字段：
+  - 26: CIViC Entity Significance (变异意义)
+  - 27: CIViC Entity Therapies (靶向药物)
+  - 28: CIViC Entity Direction (方向)
+  - 29: CIViC Entity Disease (关联疾病)
+  - 33: CIViC Evidence Level (证据等级)
+
+辅助参考字段：
+  - 34: CIViC Evidence Rating (证据评分)
+  - 35: CIViC Assertion ACMG Codes (ACMG代码)
+  - 36: CIViC Assertion AMP Category (AMP分类)
+  - 37: CIViC Assertion NCCN Guideline (NCCN指南)
+  - 38: CIViC Assertion Regulatory Approval (监管批准)
+  - 39: CIViC Assertion FDA Companion Test (FDA伴随诊断)
+"""
+
+import re
+import sys
+
+# 新字段的描述名称（按索引顺序）- 用于键值对格式
+NEW_FIELD_NAMES = [
+    "Significance",           # 26
+    "Direction",              # 27
+    "Disease",                # 28
+    "Therapies",              # 29
+    "Evidence_Level",        # 32
+    "Evidence_Rating",       # 33
+    "ACMG_Codes",            # 34
+    "AMP_Category",          # 35
+    "NCCN_Guideline",        # 36
+    "Regulatory_Approval",   # 37
+    "FDA_Companion_Test"     # 38
+]
+
+# 要保留的CSQ字段索引
+# 核心治疗相关字段（索引26-29）+ 辅助参考字段（索引32-38）
+KEEP_FIELD_INDICES = [26, 27, 28, 29, 32, 33, 34, 35, 36, 37, 38]
+
+def parse_csq_line(csq_description):
+    """从VCF header中提取CSQ字段描述"""
+    # 提取Description中的字段列表
+    match = re.search(r'Format: (.+)"', csq_description)
+    if match:
+        fields = match.group(1).split('|')
+        return fields
+    return []
+
+def filter_csq_record(csq_record, keep_indices, field_names):
+    """过滤单个CSQ记录，并转换为Key=Value格式"""
+    fields = csq_record.split('|')
+    filtered = [fields[i] for i in keep_indices if i < len(fields)]
+
+    # 转换为Key=Value格式（用分号分隔）
+    kv_pairs = []
+    for i, value in enumerate(filtered):
+        if value:  # 只添加非空值
+            kv_pairs.append(f"{field_names[i]}={value}")
+
+    return ';'.join(kv_pairs) if kv_pairs else ""
+
+def filter_csq_record(csq_record, keep_indices, field_names):
+    """将单个CSQ记录转换为Key=Value格式"""
+    fields = csq_record.split('|')
+    filtered = [fields[i] for i in keep_indices if i < len(fields)]
+
+    # 转换为Key=Value格式（用分号分隔）
+    kv_pairs = []
+    for i, value in enumerate(filtered):
+        if value:  # 只添加非空值
+            kv_pairs.append(f"{field_names[i]}={value}")
+
+    return ';'.join(kv_pairs) if kv_pairs else ""
+
+def merge_csq_records(records_string):
+    """合并多条已转换为Key=Value格式的CSQ记录，每个Key保留所有值，用|分隔"""
+    # 分割多条记录
+    records = records_string.split(',')
+
+    # 收集每个字段的所有值（不去重，按顺序保留）
+    field_values = {}
+
+    for record in records:
+        if not record:
+            continue
+        # 解析Key=Value对
+        pairs = record.split(';')
+        for pair in pairs:
+            if '=' in pair:
+                key, value = pair.split('=', 1)
+                if key not in field_values:
+                    field_values[key] = []
+                if value:
+                    field_values[key].append(value)
+
+    # 按原始字段顺序构建合并后的记录
+    merged_pairs = []
+    for name in NEW_FIELD_NAMES:
+        if name in field_values and field_values[name]:
+            # 多个值用|连接
+            merged_pairs.append(f"{name}={'|'.join(field_values[name])}")
+
+    return ';'.join(merged_pairs) if merged_pairs else ""
+
+def modify_header(info_line, keep_indices, new_field_names):
+    """修改VCF header中的INFO行"""
+    # 找到包含CSQ的INFO行
+    if 'ID=CSQ' not in info_line:
+        return info_line
+
+    # 提取Description部分 - 更精确的匹配
+    match = re.search(r'(##INFO=<ID=CSQ,Number=\.,Type=String,Description=")(.+?)(">)', info_line)
+    if not match:
+        return info_line
+
+    prefix = match.group(1)
+    old_desc = match.group(2)
+    suffix = match.group(3)
+
+    # 直接使用新的字段名
+    new_desc = '|'.join(new_field_names)
+
+    # 替换
+    new_line = prefix + new_desc + suffix
+
+    return new_line
+
+def main(input_file, output_file):
+    with open(input_file, 'r', encoding='utf-8') as f_in, \
+         open(output_file, 'w', encoding='utf-8') as f_out:
+
+        csq_fields = None
+
+        for line in f_in:
+            # 处理header行
+            if line.startswith('##'):
+                # 找到CSQ的INFO行并修改
+                if 'ID=CSQ' in line:
+                    # 先解析字段列表
+                    csq_match = re.search(r'Format: (.+)"', line)
+                    if csq_match:
+                        csq_fields = csq_match.group(1).split('|')
+
+                    # 修改header
+                    modified_line = modify_header(line.strip(), KEEP_FIELD_INDICES, NEW_FIELD_NAMES)
+                    f_out.write(modified_line + '\n')
+                else:
+                    f_out.write(line)
+            # 处理列头
+            elif line.startswith('#CHROM'):
+                f_out.write(line)
+            # 处理数据行
+            else:
+                fields = line.strip().split('\t')
+                if len(fields) >= 8:
+                    info_field = fields[7]
+
+                    # 查找并替换CSQ字段
+                    csq_match = re.search(r'CSQ=([^;]+)', info_field)
+                    if csq_match:
+                        old_csq = csq_match.group(1)
+
+                        # 处理多个CSQ记录（用逗号分隔）
+                        csq_records = old_csq.split(',')
+                        filtered_records = []
+
+                        for csq_record in csq_records:
+                            # 先转换为Key=Value格式
+                            filtered = filter_csq_record(csq_record, KEEP_FIELD_INDICES, NEW_FIELD_NAMES)
+                            if filtered:
+                                filtered_records.append(filtered)
+
+                        # 合并所有记录
+                        merged_csq = merge_csq_records(','.join(filtered_records))
+
+                        # 替换info字段中的CSQ
+                        new_info = re.sub(r'CSQ=[^;]+', f'CSQ={merged_csq}', info_field)
+                        fields[7] = new_info
+
+                    f_out.write('\t'.join(fields) + '\n')
+
+    print(f"过滤完成！输出文件: {output_file}")
+    print(f"保留的字段: {', '.join(NEW_FIELD_NAMES)}")
+
+if __name__ == '__main__':
+    input_vcf = '01-Mar-2026-civic_accepted.vcf'
+    output_vcf = 'hg19_civic.vcf'
+
+    if len(sys.argv) > 1:
+        input_vcf = sys.argv[1]
+    if len(sys.argv) > 2:
+        output_vcf = sys.argv[2]
+
+    main(input_vcf, output_vcf)