performance_profile

#!/usr/bin/env python2
import random,math
import sys, getopt, os
import time
from Bio import SeqIO

def display_help():
    print "Usage: splice_data_generator -g genome_file.fa"
    print "-g: Location of genome file"
    print "--max_size=N: integer for max size of sequence"
    print "--min_size=N: integer for minimum size of sequence"
    print "-n N: integer for number of sequences to generate"
    print "--gap N: integer for length of gap within genome between spliced sequences"

def get_detected(filename):
    f = open(filename)
    valid_chars = ['A', 'T', 'G', 'C']
    lines = 0
    for line in f:
        if line[0] in valid_chars:
            lines += 1
    return lines

def get_reads_in_file(filename):
    f = open(filename)
    lines = 0
    for line in f:
        if line[0] != '>':
            lines += 1
    return lines

positive_file = ''
negative_file = ''
genome_file = ''

try:
    opts, args = getopt.getopt(sys.argv[1:], "hp:n:g:")
except getopt.GetoptError as err:
    print 'Usage error'
    sys.exit(2)
for opt, arg in opts:
    if opt == '-p':
        positive_file = arg
    elif opt == '-h':
        display_help()
        sys.exit(2)
    elif opt == '-n':
        negative_file = arg
    elif opt == '-g':
        genome_file = arg

if positive_file == '' or negative_file == '':
    print 'Must specify positive FASTA file with -p'
    sys.exit(2)
elif negative_file == '':
    print 'Must specify negative FASTA file with -n'
    sys.exit(2)
elif genome_file == '':
    print 'Must specify genome file with -g'
    sys.exit(2)


print '--------------------------------------------------'
print 'Beginning run with positive dataset'
print '--------------------------------------------------'

begin = time.time()
os.system('./splice_detection -g ' + genome_file + ' -i ' + positive_file + ' > performance_positive_data.csv')
finish = time.time()

positive_time = finish - begin
positive_reads = get_reads_in_file(positive_file)
positive_time_per_read = positive_time/positive_reads

positive_detected = get_detected('performance_positive_data.csv')

if positive_reads != 0:
    positive_detected_percentage = (positive_detected/float(positive_reads)) * 100
else:
    positive_detected_percentage = 0

positive_time = round(positive_time, 4)
positive_time_per_read = round(positive_time_per_read, 4)
positive_detected_percentage = round(positive_detected_percentage, 3)

print '--------------------------------------------------'
print 'Ending run with positive dataset'
print '--------------------------------------------------'
print '\n'

print '--------------------------------------------------'
print 'Beginning run with negative dataset'
print '--------------------------------------------------'

begin = time.time()
os.system('./splice_detection -g ' + genome_file + ' -i ' + negative_file + ' > performance_negative_data.csv')
finish = time.time()

negative_time = finish - begin
negative_reads = get_reads_in_file(negative_file)
negative_time_per_read = negative_time/negative_reads

negative_detected = get_detected('performance_negative_data.csv')
if negative_reads != 0:
    negative_detected_percentage = (negative_detected/float(negative_reads)) * 100
else:
    negative_detected_percentage = 0

negative_time = round(negative_time, 4)
negative_time_per_read = round(negative_time_per_read, 4)
negative_detected_percentage = round(negative_detected_percentage, 3)


os.system('rm performance_negative_data.csv')
os.system('rm performance_positive_data.csv')

print '--------------------------------------------------'
print 'Ending run with negative dataset'
print '--------------------------------------------------'

print '\n\n'
print '--------------------------------------------------'
print 'Performance report'
print '--------------------------------------------------'
print 'Total time elapsed: ' + str(positive_time+negative_time) + 's'
print 'Time elapsed for positives: ' + str(positive_time) + 's'
print 'Time elapsed for negatives: ' + str(negative_time) + 's'
print 'Average time per read (positive): ' + str(positive_time_per_read) + 's'
print 'Average time per read (negative): ' + str(negative_time_per_read) + 's'
print 'Identified reads in positive dataset: ' + str(positive_detected_percentage) + '%'
print 'Identified reads in negative dataset: ' + str(negative_detected_percentage) + '%'
print '--------------------------------------------------'