过滤 MultiIndex Pandas DataFrame

Question

我有一个 MultiIndex Pandas DataFrame，如下所示：

import pandas as pd
import numpy as np

genotype_data = [
                    ['0/1', '120,60', 180, 5, 0.5, '0/1', '200,2', 202, 99, 0.01],
                    ['0/1', '200,20', 60, 99, 0.1, '0/1', '200,50', 250, 99, 0.4],
                    ['0/1', '200,2', 202, 99, 0.01, '0/1', '200,2', 202, 99, 0.01] 
]


genotype_columns = [['Sample1', 'Sample2'], ['GT', 'AD', 'DP', 'GQ', 'AB']]
cols = pd.MultiIndex.from_product(genotype_columns)
genotype = pd.DataFrame(data=genotype_data, columns=cols)

info_columns = [['INFO'], ['AC', 'DEPTH']]
cols = pd.MultiIndex.from_product(info_columns)
info = pd.DataFrame(data=[[12, 100], [23, 200], [40, 40]], columns=cols)

df = pd.concat([info, genotype], axis=1)

我想为至少有一个样本（在本例中为 Sample1 或 Sample2）具有 DP >= 50 且 GQ

我不知道从哪里开始，希望能得到一些帮助。

编辑：

在 jezrael 的帖子的帮助下，我找到了解决方案。代码如下：

genotype = df.ix[:,3:]

DP = genotype.xs('DP', axis=1, level=1)
GQ = genotype.xs('GQ', axis=1, level=1)

conditions = (DP.ge(50) & GQ.le(4)).T.any() 
df = df[conditions]

return df

Answer 1

我认为你可以使用：

• filterSamples列
• 选择xsDP和HQ子集并与ge和lt比较
• 逻辑and (&) 并通过any 获得至少一个True
• 通过idxmax 获取第一个True 并通过loc 选择

#data in sample change for matching (first 99 in HQ in Sample1 was changed to 3)

genotype_data = [
                    ['0/1', '120,60', 180, 5, 0.5, '0/1', '200,2', 202, 99, 0.01],
                    ['0/1', '200,20', 60, 3, 0.1, '0/1', '200,50', 250, 99, 0.4],
                    ['0/1', '200,2', 202, 99, 0.01, '0/1', '200,2', 202, 99, 0.01] 
]


genotype_columns = [['Sample1', 'Sample2'], ['GT', 'AD', 'DP', 'GQ', 'AB']]
cols = pd.MultiIndex.from_product(genotype_columns)
genotype = pd.DataFrame(data=genotype_data, columns=cols)

info_columns = [['INFO'], ['AC', 'DEPTH']]
cols = pd.MultiIndex.from_product(info_columns)
info = pd.DataFrame(data=[[12, 100], [23, 200], [40, 40]], columns=cols)

df = pd.concat([info, genotype], axis=1)
print (df)
  INFO       Sample1                        Sample2                       
    AC DEPTH      GT      AD   DP  GQ    AB      GT      AD   DP  GQ    AB
0   12   100     0/1  120,60  180   5  0.50     0/1   200,2  202  99  0.01
1   23   200     0/1  200,20   60   3  0.10     0/1  200,50  250  99  0.40
2   40    40     0/1   200,2  202  99  0.01     0/1   200,2  202  99  0.01

df1 = df.filter(like='Sample')
df = df.loc[[(df1.xs('DP', axis=1, level=1).ge(50) & 
              df1.xs('GQ', axis=1, level=1).lt(4)).any(1).idxmax()]]
print (df)
  INFO       Sample1                     Sample2                      
    AC DEPTH      GT      AD  DP GQ   AB      GT      AD   DP  GQ   AB
1   23   200     0/1  200,20  60  3  0.1     0/1  200,50  250  99  0.4

编辑：

如果需要按条件返回所有行，删除loc和idmax：

df1 = df.filter(like='Sample')
#changed condition to lt(10) (<10)
df = df[(df1.xs('DP', axis=1, level=1).ge(50) & df1.xs('GQ', axis=1, level=1).lt(10)).any(1)]
print (df)
  INFO       Sample1                      Sample2                       
    AC DEPTH      GT      AD   DP GQ   AB      GT      AD   DP  GQ    AB
0   12   100     0/1  120,60  180  5  0.5     0/1   200,2  202  99  0.01
1   23   200     0/1  200,20   60  3  0.1     0/1  200,50  250  99  0.40

Answer 2

stack第一级，使用query标识索引

df.loc[df.stack(0).query('DP >= 50 & GQ < 4').unstack().index]

  INFO       Sample1                     Sample2                      
    AC DEPTH      GT      AD  DP GQ   AB      GT      AD   DP  GQ   AB
1   23   200     0/1  200,20  60  3  0.1     0/1  200,50  250  99  0.4

我使用了@jezrael 的设置

genotype_data = [
                    ['0/1', '120,60', 180, 5, 0.5, '0/1', '200,2', 202, 99, 0.01],
                    ['0/1', '200,20', 60, 3, 0.1, '0/1', '200,50', 250, 99, 0.4],
                    ['0/1', '200,2', 202, 99, 0.01, '0/1', '200,2', 202, 99, 0.01] 
]


genotype_columns = [['Sample1', 'Sample2'], ['GT', 'AD', 'DP', 'GQ', 'AB']]
cols = pd.MultiIndex.from_product(genotype_columns)
genotype = pd.DataFrame(data=genotype_data, columns=cols)

info_columns = [['INFO'], ['AC', 'DEPTH']]
cols = pd.MultiIndex.from_product(info_columns)
info = pd.DataFrame(data=[[12, 100], [23, 200], [40, 40]], columns=cols)

df = pd.concat([info, genotype], axis=1)