import argparse
import mcb185
import statistics
parser = argparse.ArgumentParser(desciption='stats about sequence')
#required arguments
parser.add_argument('--file',
required=True,
type=str,
metavar='<str>',
help='required fasta file')
arg = parser.parse_args()
length = []
for name, seq in mcb185.read_fasta(arg.file):
#print(name, len(seq))
length.append(len(seq))
length.sort()
#print(length)
#MIN:
print('min is', min(length))
#MAX:
print('max is', man(length))
#sum:
#sum = 0
#for value in length:
# sum += value
# Make a program that reports the amino acid composition in a file of proteins
import mcb185
import sys
total = 0
count = {}
for name, seq in mcb185.read_fasta(sys.argv[1]):
for aa in seq:
if aa == "*":
continue
if aa in count:
count[aa] += 1
else:
count[aa] = 1
total += 1
for aa in count:
print(aa, count[aa], count[aa] / total)
"""
python3 aa_comp.py -- fasta at_prots.fa | sort -nk2
W 528 0.012054244098442994
C 801 0.018286836217524315
H 1041 0.023766038080452946
M 1097 0.025044518515136296
Y 1281 0.02924523994338158
Q 1509 0.03445048171316378
F 1842 0.04205287429797726
N 1884 0.04301173462398977
P 2051 0.046824345920277614
T 2153 0.04915300671202228
R 2320 0.05296561800831012
import sys
import argparse
import mcb185
parser = argparse.ArgumentParser(description = 'translate RNAs to proteins')
parser.add_argument('--seq', required = True, type = str, nargs = '+', metavar = '<str>',
help = 'DNA sequ (file or sequences sep by spaces))')
arg = parser.parse_args()
if '.fa' in arg.seq:
for name, seq in mcb185.read_fasta(arg.seq):
print(f'>{name}') # Will print >name - looks like a fasta file
print(mcb185.translate(seq))
else:
for string in arg.seq:
print(mcb185.translate(string))
"""
for seq in sys.argv[1:]:
prot = ''
for i in range(0, len(seq), 3):
prot += gcode[seq[i:i+3]]
print(prot)
"""
# You have been given the code above
# An example command line is
# python3 translate.py ATGCGCCCGAACTAG ATGAAACCCGGGTTT
# Your task is to write a new program with the following features
# 1. Proper command line (argparse)
# 2. Reads sequences in from fasta format rather than sys.argv
