Source code for jass_preprocessing.map_gwas

"""
Map GWAS

A set of functions to find GWAS files in subfolder and
to map columns

"""

import os
import sys
import pandas as pd
import numpy as np

[docs]def walkfs(startdir, findfile): """ Go through the folder and subfolder to find the specified file Args: startdir (str): path of the folder to explore findfile (str): name of the file to find """ dircount = 0 filecount = 0 for root, dirs, files in os.walk(startdir): if findfile in files: return dircount, filecount + files.index(findfile), os.path.join(root, findfile) dircount += 1 filecount += len(files) # nothing found, return None instead of a full path for the file return dircount, filecount, None
[docs]def convert_missing_values(df): """ Convert all missing value strings to a standart np.nan value Args: GWAS_table (pandas dataframe): GWAS data as a dataframe Return: a pandas dataframe with missing value all equal to np.nan """ def_missing = ['', '#N/A', '#N/A', 'N/A', '#NA', '-1.#IND', '-1.#QNAN', '-NaN', '-nan', '1.#IND', '1.#QNAN', 'N/A', 'NA', 'NULL', 'NaN', 'nan', 'na', '.'] nmissing = len(def_missing) nan_vec = ["NA"] * nmissing return df.replace(def_missing, nan_vec)
[docs]def map_columns_position(gwas_internal_link, GWAS_labels): """ Find column position for each specific Gwas Args: gwas_internal_link (str): filename of the GWAS data (with path) GWAS_labels (str): filename of the csv information file Return: pandas Series with column position and column names as index """ column_dict = pd.read_csv(GWAS_labels, sep='\t', na_values='na', index_col=0) gwas_file = gwas_internal_link.split('/')[-1] my_labels = column_dict.loc[gwas_file] #Our standart labels: reference_label = column_dict.columns.tolist() # labels in the GWAS files target_lab = pd.Index(my_labels.values.tolist()) f = open(gwas_internal_link) count_line = 0 line = f.readline() print(line) header = pd.Index(line.split()) def get_position(I,x): try: return I.get_loc(x) except KeyError: return np.nan label_position = [get_position(header,i) for i in target_lab] mapgw = pd.Series(label_position, index=reference_label) mapgw = mapgw.loc[~mapgw.isna()].astype(int) mapgw.sort_values(inplace=True) print(mapgw) f.close() return mapgw
[docs]def read_gwas( gwas_internal_link, column_map): """ Read gwas raw data, fetch columns thanks to position stored in column_map and rename columns according to column_map.index Args: gwas_internal_link (str): GWAS data as a dataframe column_map (pandas Series): Series containing the position of column in the raw data Return: a pandas dataframe with missing value all equal to np.nan """ fullGWAS = pd.read_csv(gwas_internal_link, delim_whitespace=True, usecols = column_map.values, #column_dict['label_position'].keys(), names= column_map.index, index_col=0, header=0, na_values= ['', '#N/A', '#N/A', 'N/A', '#NA', '-1.#IND', '-1.#QNAN', '-NaN', '-nan', '1.#IND', '1.#QNAN', 'N/A', 'NA', 'NULL', 'NaN', 'nan', 'na', '.']) fullGWAS = fullGWAS[~fullGWAS.index.duplicated(keep='first')] #fullGWAS = convert_missing_values(fullGWAS) return fullGWAS