def prep(pid):
"""
Selects expert move sets and saves in npy file
:param pid: Puzzle ID
:return: npy saved files of training data
"""
# pidList = pid
# pid = pidList[0]
threshold = 50
print 'ready with pid %i' % pid
data,users,encoded_bf,lens = experience_labs(pid,threshold)
print 'experience_labs'
encoded = encode_movesets_style_pr(data)
encoded_base = encode_bases(data)
encoded_loc = encode_location(data,len_longest)
print 'encoded'
# plist = []
# lens = []
# for pid in pidList:
# puzzles_pid = (moveset_dataFrame.loc[moveset_dataFrame['pid'] == pid])
# for uid in users:
# puzzles_pid2 = puzzles_pid.loc[puzzles_pid['uid'] == uid]
# p = (list(puzzles_pid2['move_set']))
# plist.extend(p)
# lens.append(len(list(puzzles_pid['move_set'])))
#
# bf_list = []
# for i in plist:
# s1 = (ast.literal_eval(i))
# s2 = s1['begin_from']
# bf_list.append(s2)
#
# encoded_bf = []
# for start in bf_list:
# enc = []
# for i in start:
# if i == 'A':
# enc.append(1)
# elif i == 'U':
# enc.append(2)
# elif i == 'G':
# enc.append(3)
# elif i == 'C':
# enc.append(4)
# encoded_bf.append(enc)
print 'encoded_bf'
print len(encoded), len(encoded_bf), len(data)
print lens
bases = base_sequence_at_current_time_pr(encoded,encoded_bf)
#bases = base_sequence_at_current_time_pr(encoded[1006],encoded_bf[1006])
X = np.array(structure_and_energy_at_current_time(bases,pid,data,len_longest))
np.save(open(os.getcwd()+'/npsaves/X-exp-'+str(pid),'wb'),X)
np.save(open(os.getcwd()+'/npsaves/y-exp-base-'+str(pid),'wb'),encoded_base)
np.save(open(os.getcwd()+'/npsaves/y-exp-loc-'+str(pid),'wb'),encoded_loc)
def read(pid, uidList):
"""
Returns training data for expert players of one puzzle
:param pid: Puzzle ID
:param uidList: List of user IDs
:return: Pickled training data
"""
#print 'ready with pid %i' % pid
#uidList.remove(87216)
#uidList = [8627]
#print uidList
final_dict = []
bf_list = []
#start = time.time()
for user in uidList:
#print user
data = read_movesets_uid_pid(user, pid)
#data = read_movesets_uid(user)
#print 'data read'
if not data:
#print 'user %i with pid %i list empty' % (user,pid)
continue
else:
for i in data:
#print 'formatting into list'
s1 = ast.literal_eval(i)
s2 = s1['moves']
s3 = s1['begin_from']
final_dict.append(s2)
bf_list.append(s3)
#print 'done formatting list'
#print 'user %i done with pid %i' % (user,pid)
##print time.time() - start()
#print "complete data read"
encoded_bf = []
for start in bf_list:
enc = []
for i in start:
if i == 'A':
enc.append(1)
elif i == 'U':
enc.append(2)
elif i == 'G':
enc.append(3)
elif i == 'C':
enc.append(4)
encoded_bf.append(enc)
#print "encoded begin_from"
encoded = encode_movesets_style_pr(final_dict)
encoded_base = (encode_bases(final_dict))
encoded_loc = (encode_location(final_dict, len_longest))
#print 'encoded base and location'
#print len(encoded), len(encoded_bf), len(final_dict)
bases = base_sequence_at_current_time_pr(encoded, encoded_bf)
#print 'encoded base seqs'
#bases = base_sequence_at_current_time_pr(encoded[1006],encoded_bf[1006])
X = (structure_and_energy_at_current_time(bases, pid))
#X2 = (structure_and_energy_at_current_time_with_location(bases,pid,final_dict,len_longest))
#print 'encoded strucs energy and locks'
#print len(X)
# np.save(open(os.getcwd()+'/npsaves/X-exp-base-eli.npy','wb'),X2)
# np.save(open(os.getcwd()+'/npsaves/X-exp-loc-eli.npy','wb'),X)
# np.save(open(os.getcwd()+'/npsaves/y-exp-base-eli.npy','wb'),encoded_base)
# np.save(open(os.getcwd()+'/npsaves/y-exp-loc-eli.npy','wb'),encoded_loc)
#pickle.dump(X2,open(os.getcwd()+'/pickles/X-exp-base-'+str(pid),'wb'))
pickle.dump(X, open(os.getcwd() + '/pickles/X-hog-loc-' + str(pid), 'wb'))
pickle.dump(encoded_base,
open(os.getcwd() + '/pickles/y-hog-base-' + str(pid), 'wb'))
pickle.dump(encoded_loc,
open(os.getcwd() + '/pickles/y-hog-loc-' + str(pid), 'wb'))
def read_uid(uidList):
"""
Returns training data for expert players of one puzzle
:param pid: Puzzle ID
:param uidList: List of user IDs
:return: Pickled training data
"""
#print 'ready with pid %i' % pid
#uidList.remove(87216)
#uidList = [8627]
#print uidList
final_dict = []
bf_list = []
count = 0
#start = time.time()
for user in uidList:
#print user
data = read_movesets_uid(user)
#data = read_movesets_uid(user)
#print 'data read'
if not data:
#print 'user %i with pid %i list empty' % (user,pid)
continue
else:
for i in data:
#print 'formatting into list'
try:
s1 = ast.literal_eval(i)
s2 = s1['moves']
s3 = s1['begin_from']
final_dict.append(s2)
bf_list.append(s3)
print(s2, s3)
count += len(s2)
except:
continue
#print 'done formatting list'
print('Comnpleted %i/%i' % (uidList.index(user), len(uidList)))
##print time.time() - start()
#print "complete data read"
encoded_bf = []
for start in bf_list:
enc = []
for i in start:
if i == 'A':
enc.append(1)
elif i == 'U':
enc.append(2)
elif i == 'G':
enc.append(3)
elif i == 'C':
enc.append(4)
encoded_bf.append(enc)
#print "encoded begin_from"
encoded = encode_movesets_style_pr(final_dict)
encoded_base = (encode_bases(final_dict))
encoded_loc = (encode_location(final_dict, len_longest))
bases = base_sequence_at_current_time_pr(encoded, encoded_bf)
X = (structure_and_energy_at_current_time(bases, pid))
return X
def speed(pid):
"""
Encodes the puzzle solutions that were completed in the fewest
number of moves
:param pid: Puzzle ID
:return: CNN training data of fastest solutions for that puzzle
"""
final_dict = []
bf_list = []
#for pid in pidList:
print(pid)
puzzles_pid = moveset_dataFrame.loc[moveset_dataFrame['pid'] == pid]
plist = list(puzzles_pid['move_set'])
ulist = list(puzzles_pid['uid'])
plist_dict = []
for i in (plist):
s1 = (ast.literal_eval(i))
s2 = int(s1['num_moves'])
if s2 <= max_moves: # solved in 50 moves or less
print('fast')
s3 = s1['moves']
s4 = s1['begin_from']
final_dict.append(s3)
bf_list.append(s4)
else:
continue
print("complete data read")
encoded_bf = []
for start in bf_list:
enc = []
for i in start:
if i == 'A':
enc.append(1)
elif i == 'U':
enc.append(2)
elif i == 'G':
enc.append(3)
elif i == 'C':
enc.append(4)
encoded_bf.append(enc)
print("encoded begin_from")
print(len(final_dict))
encoded = encode_movesets_style_pr(final_dict)
encoded_base = (encode_bases(final_dict))
encoded_loc = (encode_location(final_dict, len_longest))
print('encoded base and location')
print(len(encoded), len(encoded_bf), len(final_dict))
bases = base_sequence_at_current_time_pr(encoded, encoded_bf)
print('encoded base seqs')
#print len(bases[0][0])
#bases = base_sequence_at_current_time_pr(encoded[1006],encoded_bf[1006])
X = (structure_and_energy_at_current_time(bases, pid))
#X2 = (structure_and_energy_at_current_time_with_location(bases,pid,final_dict,len_longest))
print('encoded strucs energy and locks')
print(len(X))
# np.save(open(os.getcwd()+'/npsaves/X-exp-base-eli.npy','wb'),X2)
# np.save(open(os.getcwd()+'/npsaves/X-exp-loc-eli.npy','wb'),X)
# np.save(open(os.getcwd()+'/npsaves/y-exp-base-eli.npy','wb'),encoded_base)
# np.save(open(os.getcwd()+'/npsaves/y-exp-loc-eli.npy','wb'),encoded_loc)
#pickle.dump(X2,open(os.getcwd()+'/pickles/X-exp-base-'+str(pid),'wb'))
if len(encoded) != 0:
pickle.dump(
X, open(os.getcwd() + '/pickles/X2-fast-loc-' + str(pid), 'wb'))
pickle.dump(
encoded_base,
open(os.getcwd() + '/pickles/y2-fast-base-' + str(pid), 'wb'))
pickle.dump(
encoded_loc,
open(os.getcwd() + '/pickles/y2-fast-loc-' + str(pid), 'wb'))
from encodeRNA import encode_movesets_style, base_sequence_at_current_time, structure_and_energy_at_current_time from encodeRNA import encode_bases, base_sequence_at_current_time_pr, encode_movesets_style_pr, encode_location import numpy as np import pandas as pd import ast import copy import pickle pid = 6502997 len_puzzle = 80 filepath = os.getcwd() + '/movesets/moveset6-22a.txt' data2, users = read_movesets_pid(filepath,pid) data = data2 encoded = (encode_movesets_style_pr(data)) encoded_base = encode_bases(data) encoded_loc = encode_location(data,len_puzzle) moveset_dataFrame = pd.read_csv(filepath, sep=" ", header="infer", delimiter='\t') puzzles_pid = (moveset_dataFrame.loc[moveset_dataFrame['pid'] == pid]) structure_file = os.getcwd() + '/movesets/puzzle-structure-data.txt' #print puzzles_pid plist = list(puzzles_pid['move_set']) #print plist bf_list = [] for i in plist: s1 = (ast.literal_eval(i)) s2 = s1['begin_from'] bf_list.append(s2)