602panel_test/igv_input.py

import pandas as pd
import os
import argparse

def igv_format(inputpath,sample):
	##0.creat the analysis path
	tempfile = os.path.join(inputpath, 'tempfile')
	if not os.path.exists(tempfile):
		os.mkdir(tempfile)
	igv = os.path.join(tempfile, 'igv')
	if not os.path.exists(igv):
		os.mkdir(igv)
	#1.acqure the dir
	laneid = inputpath.split('/')[-1]
	resultdir = os.path.join(inputpath, 'resultfile')
	sampledir = os.path.join(resultdir, sample)
	#2for somatic sv
	snvdir = os.path.join(sampledir, laneid + '-' + sample + '.snv.xlsx')
	snv_raw = pd.read_excel(snvdir)
	select_columns = ['Chr', 'Start', 'End', 'Ref', 'Alt', 'Gene.refGene', 'Transid_uniq','ExonicFunc.refGene', 'VAF_tumor', 'DP_tumor',
				  'VD_tumor', 'DP_normal', 'VD_normal', 'VAF_normal', 'sampleid']
	snv_select = snv_raw[select_columns]
	snv_select['Chr'] = 'chr' + snv_select['Chr'].astype('str')
	snvoutputname = os.path.join(igv, laneid + '-' + sample + '.snv.txt')
	snv_select.to_csv(snvoutputname, sep='\t', index=False, header=True)
	#3.for germline sv
	germlinedir = os.path.join(sampledir, laneid + '-' + sample + '.germline.xlsx')
	germline_raw = pd.read_excel(germlinedir)
	germline_columns = ['Chr', 'Start', 'End', 'Ref', 'Alt', 'Gene.refGene', 'AAChange_select', 'ExonicFunc.refGene','VAF','N_ref','N_alt','sampleid']
	germline_select = germline_raw[germline_columns]
	germline_select['Chr'] = 'chr' + germline_select['Chr'].astype('str')
	germlineoutputname = os.path.join(igv, laneid + '-' + sample + '.germline.txt')
	germline_select.to_csv(germlineoutputname, sep='\t', index=False, header=True)





def allsample(inputpath):
	laneid = inputpath.split('/')[-1]
	sampledir = os.path.join(inputpath, laneid + '_sample_infor_label.txt')
	samplelist = pd.read_table(sampledir, sep='\t', header=0)
	for i in range(len(samplelist)):
		sample = samplelist.loc[i, 'samplename']
		igv_format(inputpath, sample)



if __name__=='__main__':
	parser = argparse.ArgumentParser(description='the igv for sv')
	parser.add_argument('-i', '--inputpath', type=str, help='the path of lane')
	parser.add_argument('-s', '--sample', type=str, help='the samplename')
	args = parser.parse_args()
	Inputpath = args.inputpath
	Sample=args.sample
	igv_format(Inputpath,Sample)