IGEM:Harvard/2006/Container Design 4/Python Code

Script for adding aptamers:

#######
# Add aptamers to the ends of the appropriate oligos.
#######

# Constants
apt_seq = 'GGTTGGTGTGGTTGG'
T_linker = 'TTT'
        
print oligo_ra
num_aptamers = int(raw_input('How many aptamers do you want to add? '))
i = 0   
while i < num_aptamers:
        oligo_num = int(raw_input('Which oligo needs an aptamer? '))
        if oligo_num < len(oligo_ra):
                # Add the aptamer to that oligo
                oligo_ra[oligo_num] = oligo_ra[oligo_num] + T_linker + apt_seq
                i = i + 1
        else:
                print 'oligo ' + str(oligo_num) + ' out of range.'

print oligo_ra

Add aptamers using File Input rather than User Input

#####
# Add apts this time using file input instead of user input
#####
                        
# Constants
apt_seq = 'GGTTGGTGTGGTTGG'
T_linker = 'TTT'
                        
fin_barrel = None
                        
try:
        fin_barrel = open("barrel_apts_to_add.txt", "r")
except IOError, e:
        print "Error in file IO: ", e

# Ask the user if they are running a lid or a barrel
shape = int(raw_input("Enter 1 if you are running a barrel, 2 if lid: "))
if (shape == 1):   
        apts_to_add = pickle.load(fin_barrel)
        for apt_specs in apts_to_add:
                oligo_num = apt_specs[0]
                type = apt_specs[1]
                if (type == 1):
                        # apt is pointing in so add 'I' as a flag at the end
                        oligo_ra[oligo_num] = oligo_ra[oligo_num] + T_linker + apt_seq + 'I'
                elif(type == 2):
                        # apt is pointing out so add 'O' as a flag
                        oligo_ra[oligo_num] = oligo_ra[oligo_num] + T_linker + apt_seq + 'O'
                else:
                        # incorrect type
                        print 'Bad input - aptamer needs to be pointing in or out'

Script for splitting up as many oligos as you want (USING USER INPUT). The first part goes in main. The second part goes in honeycomb_pointers_v1.py

#####
# Oligo splitting:
# get new list of oligos given user input specifying which oligo to cut
#####

num_to_split = int(raw_input('How many oligos do you want to split?'))
i = 0
new_OTP_ra = OTP_ra[:]
while i < num_to_split:  

        oligo_num = int(raw_input('Enter the number oligo you wish to split:'))
        print '\n'
        print 'How many tokens should the first new oligo be?'
        num_toks = int(raw_input('Number of toks starting from 5 prime: '))

        new_OTP_ra = split_oligo(new_OTP_ra, oligo_num, num_toks)
        
        print new_OTP_ra
        print len(OTP_ra)
        print len(new_OTP_ra)
        i = i + 1

####
# given an oligo to split, split it and return the new list of oligos
####    
def split_oligo(new_OTP_ra, oligo_num, num_toks):
        
        print new_OTP_ra[oligo_num]

        original_oligo = new_OTP_ra[oligo_num]
        
        oligo_1 = original_oligo[:num_toks]  
        oligo_2 = original_oligo[num_toks:]
        print oligo_1
        print'\n'
        print oligo_2
                
        new_OTP_ra[oligo_num] = oligo_1
        new_OTP_ra.insert(oligo_num + 1, oligo_2)
        return new_OTP_ra    

Script for splitting up oligos automatically (FILE INPUT). (make sure to import pickle at the top). This part goes in main. Further down is the file to read from essentially (uses pickle)

#####
# Oligo splitting - this time reading from a file and not asking for user
# input
#####
fin_barrel = None
fin_lid = None
        
try:
        fin_barrel = open("barrel_oligos_to_split.txt", "r")
        fin_lid = open("lid_oligos_to_split.txt", "r")
except IOError, e:
        print "Error in file IO: ", e
        
# Ask the user if they are running a lid or a barrel
shape = int(raw_input("Enter 1 if you are running a barrel, 2 if lid: "))
if (shape == 1):
        oligos_to_split = pickle.load(fin_barrel)
elif (shape == 2):
        oligos_to_split = pickle.load(fin_lid)
else:
        print 'Please modify code or run with lid or 30hb barrel'


new_OTP_ra = OTP_ra[:]
for pair in oligos_to_split:
        oligo_num = pair[0]
        print oligo_num
        num_toks = pair[1]
        print num_toks

        new_OTP_ra = split_oligo(new_OTP_ra, oligo_num, num_toks)
        
        print new_OTP_ra
        print len(OTP_ra)
        print len(new_OTP_ra)

# if it's the barrel design and so that all the numbers aren't messed
# up, the 7bp token on the start of strand 10 needs to be removed
# because it's going to be left unpaired
if (shape == 1):
        new_OTP_ra = new_OTP_ra[:61] + new_OTP_ra[62:]

if fin_barrel: fin_barrel.close()
if fin_lid: fin_lid.close() 
        

### split = [[oligo_num, num_tokens for oligo_1 - from 5prime], ... ]

import pickle

fout_barrel = None
fout_lid = None

try:
        fout_barrel = open("barrel_oligos_to_split.txt", "w")
        fout_lid = open("lid_oligos_to_split.txt", "w")
except IOError, e:
        print "Error in file IO: ", e

barrel_split = [[56, 2], [57, 3], [41, 3], [26, 3], [21, 4]]
lid_split = [[27, 4], [2, 4]]

pickle.dump(barrel_split, fout_barrel)
pickle.dump(lid_split, fout_lid)

# clean up if they're open
if fout_barrel:
        fout_barrel.close()
if fout_lid:
        fout_lid.close()

Script for generating sequence chunks

#!/usr/bin/python

import string
import sys
import fileinput

def nowhite(s):
  return ''.join([c for c in s if c in string.letters])

seq = ''

# read in sequence
for line in fileinput.input("-"):
  seq = seq + nowhite(line)

lid_1 = 96 * 13
linker_1 = 30
box = 29 * 96
linker_2 = 30
lid_2 = 96 * 14
linker_3 = 200
box_2 = 96

#store first segment up to end of lid 1
lid_1_seq = seq[:lid_1]
#reset sequence to start at end of lid 1
seq = seq[lid_1:]
linker_1_seq = seq[:linker_1]
seq = seq[linker_1:]
box_1_seq = seq[:box]
seq = seq[box:]
linker_2_seq = seq[:linker_2]
seq = seq[linker_2:]
lid_2_seq = seq[:lid_2]
seq = seq[lid_2:]
linker_3_seq = seq[:linker_3]
seq = seq[linker_3:]
box_2_seq = seq[:box_2]
seq = seq[box_2:]
extra_scaffold = seq
barrel = box_1_seq + box_2_seq

#print out the length of each segment and its sequence

barrel_file = file('barrel_scaffold.txt', 'w')
lid_1_file = file('lid_1_scaffold.txt', 'w')
lid_2_file = file('lid_2_scaffold.txt', 'w')

print len(lid_1_seq)
print lid_1_seq
lid_1_file.write(lid_1_seq)
print len(linker_1_seq)
print linker_1_seq
print len(barrel)
print barrel
barrel_file.write(barrel)
print len(linker_2_seq)
print linker_2_seq  
print len(lid_2_seq)
print lid_2_seq
lid_2_file.write(lid_2_seq)
print len(linker_3_seq)
print linker_3_seq

Oligo sorting - find an print out those with aptamers

#####   
# oligo sorting
#####

# sort based on whether or not there's an aptamer attached to the end of
# an oligo
        
apt = re.compile('TTTGGTTGGTGTGGTTGG')
oligo_num = 0
for oligo in oligo_ra:   
        m = apt.search(oligo)
        if m:
                print 'Match found: ', oligo + ' : ' + str(oligo_num)
        else: 
                print 'No match' + str(oligo_num)
        oligo_num = oligo_num + 1

Modifications to William's program to print each oligo number next to what tokens it represents

oligo_num = 0
for oligo in OTP_ra:
        for token in oligo:
                print str(oligo_num) + ": ", token
        oligo_num = oligo_num + 1

Modifications to William's program to print a grid of oligo numbers completely filled in

• add this part to main (AAA or BBB)

####
# generate and print the oligo grid
####

# Initialize the grid with all periods
num_strands = len(TPP_ra)
num_subzones = len(TPP_ra[0])

sub_token_visit_ra = ['.' for subzone_num in range(num_subzones)]
grid_ra = [sub_token_visit_ra[:] for strand_num in range(num_strands)]
        
oligo_num = 0
for oligo in OTP_ra:
        grid_ra = generate_oligo_path(oligo, oligo_num, grid_ra)
        oligo_num = oligo_num + 1
print grid_ra
        
print_all_oligos(grid_ra, num_strands, num_subzones)

• add this part to honeycomb_pointers_v1.py

# The idea here is to have a function that adds the numbers of one oligo path
# to the appropriate places in the big grid array. Eventually this will be printed
# in main. Also it needs to be initialized in main. Oligo_path is the path of
# one oligo, while grid_ra is the grid that is constantly being updated until 
# it is printed in main. oligo_num is number that will be inputed to the grid_ra.
                        
def generate_oligo_path(oligo_path, oligo_num, grid_ra):
        num_path_tokens = len(oligo_path)
                
# Assign visits
        for path_token_num in range(num_path_tokens):
                token = oligo_path[path_token_num]
                strand = token[0]
                subzone = token[1]
                grid_ra[strand][subzone] = oligo_num
        
        
        return grid_ra

def print_all_oligos(grid_ra, num_strands, num_subzones):
        spacer = '   '
        for strand_num in range(num_strands):
                for subzone_num in range(num_subzones):
                        visitor_string = str(grid_ra[strand_num][subzone_num])   
                        sys.stdout.write(visitor_string)
                        sys.stdout.write(spacer[:4 - len(visitor_string)])
                sys.stdout.write('\n')