clean_data.py

import sys
import pandas as pd


file_names = [sys.argv[1], sys.argv[2]]

frames = []
output_filename = ""

for file_name in file_names:
	df = pd.read_csv(file_name, sep='\t')
	df = df[['nucleotides', 'phylum_name']]
	df = df[~df.nucleotides.str.contains('-')]
	df = df[:10]	

	df.columns = ['X', 'Y']
	frames.append(df)
	output_filename += (file_name[:-4] + "_")
	print(df)

print(frames)
data = pd.concat(frames)
data.to_csv(output_filename + "out.csv", index=False)
print(df)