shawnlyu / quantile_normalize Goto Github PK

Dear ShawnLYU,

nice piece of code! what you think about preserving np.nan values in the dataframe?
cheers

Michele

It says in the 'Readme' file:

Input data is a Pandas dataframe (df). Each row stands for an observations and each column stands for an attribute

Not sure if I got it right with what you meant by 'observations' and 'attribute'. According to Quantile Normalization, in the case of gene expression data, your codes are actually fit for input with each row being a gene (attribute?) and each column being a sample (observation?).

I found the code useful. I rewrote the function using 2D numpy array as input in my use case. I'm posting it here if it's of any help to others or as an addition to your code.

import numpy as np
import pandas as pd

def qnorm_np(arr):
    sorted_arr = arr.astype(float)
    sorted_arr.sort(axis=0)
    rank = sorted_arr.mean(axis=1)
    for col in range(arr.shape[1]):
        t = np.searchsorted(np.sort(arr[:, col]), arr[:,col])
        sorted_arr[:,col] = rank[t]
    return sorted_arr

#Test
df_input =  pd.DataFrame({'C1': {'A': 5, 'B': 2, 'C': 3, 'D': 4},
                   'C2': {'A': 4, 'B': 1, 'C': 4, 'D': 2},
                   'C3': {'A': 3, 'B': 4, 'C': 6, 'D': 8}})
arr_norm = qnorm_np(df_input.values)

• P