生信技能29 - 提取VCF文件测序深度和基因型信息

提取VCF文件测序深度和基因型信息，示例如下，结果保存至excel文件中。

程序代码

import os
import sys
import openpyxl
import pandas as pd

def get_depth_genetpye(vcf_path, output_path):
    print("read raw vcf...")
    # 越过50行读取VCF文件
    df = pd.read_csv(vcf_path, sep='\t', skiprows=50)
    # 提取vcf文件的染色体编号、染色体位置和INFO列数据
    df_filter = df.iloc[:, [0,1,-1]]
    # 获取样本名称/样本编号
    sample = list(df_filter.columns)[-1]

    for idx, row in df_filter.iterrows():
        info = row[sample]
        if ':' in info:
            list_info = info.split(':') 
            # 位点测序深度
            df_filter.loc[idx, 'Site_depth'] = ':'.join([list_info[1], list_info[2]])
            # 测序深度
            df_filter.loc[idx, 'Depth'] = ':'.join([list_info[1], list_info[2]])
					 # 基因型
            df_filter.loc[idx,'Genetype'] = str(list_info[0])
        else:
        		 df_filter.loc[idx, 'Site_depth'] = 'ERROR'
            df_filter.loc[idx, 'Depth'] = 'ERROR'
            df_filter.loc[idx, 'Genetype'] = 'ERROR'
    df_filter = df_filter[['#CHROM', 'POS', 'Depth', 'Genetype']]
    # 更改列名
    df_filter.columns = ['Chr', 'Start', 'Depth', 'Genetype']
    # 保存结果
    df_filter.to_excel(output_path, index=False)

结果文件