damn_son

from keras.models import Sequential
from keras.layers import Dense
from keras.layers import LSTM
from keras import losses
import pandas as pd
import numpy as np
from numpy import argmax
import matplotlib.pyplot as plt
import time

prop = {1:[1.8,-0.17,0.11,0,0.38,-0.21,-1.6,0.42,-0.27,1.12,0.61],
18:[-4.5,-0.81,2.58,3.71,-2.57,2.11,12.3,-1.56,1.87,-2.55,0.6],
14:[-3.5,-0.42,2.05,3.47,-1.62,0.96,4.8,-1.03,0.81,-0.83,0.06],
4:[-3.5,-1.23,3.49,2.95,-3.27,1.36,9.2,-0.51,0.81,-0.83,0.46],
3:[2.5,0.24,-0.13,0.49,-0.3,-6.04,-2,0.84,-1.05,0.59,1.07],
17:[-3.5,-0.58,2.36,3.01,-1.84,1.52,4.1,-0.96,1.1,-0.78,0],
5:[-3.5,-2.02,2.68,1.64,-2.9,2.3,8.2,-0.37,1.17,-0.92,0.47],
7:[-0.4,-0.01,0.74,1.72,-0.19,0,-1,0,-0.16,1.2,0.07],
8:[-3.2,-0.96,2.06,4.76,-1.44,-1.23,3,-2.28,0.28,-0.93,0.61],
9:[4.5,0.31,-0.6,-1.56,1.97,-4.81,-3.1,1.81,-0.77,1.16,2.22],
12:[3.8,0.56,-0.55,-1.81,1.82,-4.68,-2.8,1.8,-1.1,1.18,1.53],
11:[-3.9,-0.99,2.71,5.39,-3.46,3.88,8.8,-2.03,1.7,-0.8,1.15],
13:[1.9,0.23,-0.1,-0.76,1.4,-3.66,-3.4,1.18,-0.73,0.55,1.18],
6:[2.8,1.13,-0.32,-2.2,1.98,-4.65,-3.7,1.74,-1.43,0.67,2.02],
16:[-1.6,-0.45,2.23,-1.52,-1.44,0.75,0.2,0.86,-0.75,0.54,1.95],
19:[-0.8,-0.13,0.84,1.83,-0.53,1.74,-0.6,-0.64,0.42,-0.05,0.05],
20:[-0.7,-0.14,0.52,1.78,-0.32,0.78,-1.2,-0.26,0.63,-0.02,0.05],
23:[-0.9,1.85,0.3,-0.38,1.53,-3.32,-1.9,1.46,-1.57,-0.19,2.65],
25:[-1.3,0.94,0.68,-1.09,0.49,-1.01,0.7,0.51,-0.56,-0.23,1.88],
22:[4.2,-0.07,-0.31,-0.78,1.46,-3.5,-2.6,1.34,-0.4,1.13,1.32],
26:[0,0,0,0,0,0,0,0,0,0,0]}

df_train=np.asarray(pd.read_csv('https://raw.githubusercontent.com/sak1b0/proteiNN/master/train_formatted.csv',header=None))
df_test=np.asarray(pd.read_csv('https://raw.githubusercontent.com/sak1b0/proteiNN/master/test_formatted.csv',header=None))

x_train = df_train[:,0]
y_train = df_train[:,1]

x_test = df_test[:,0]
y_test = df_test[:,1]

def debug_me():
  #print('train dataframe: ',df_train.shape)
  print('x_train shape: ',x_train.shape)
  print('y_train shape: ',y_train.shape)

  #print('test dataframe: ',df_test.shape)
  print('x_test shape: ',x_test.shape)
  print('y_test shape: ',y_test.shape)

max_len=400

#================== x_train ===============
n = x_train
j=-1

for i in x_train:
  j=j+1
  if(len(i)>max_len):
    n = np.delete(n, j)
    j=j-1

for item in range (len(n)):
  n[item] = n[item]+'Z'*(max_len-len(n[item]))

x_train = n

#================= y_train =================
n = y_train
j=-1

for i in y_train:
  j=j+1
  if(len(i)>max_len):
    n = np.delete(n, j)
    j=j-1

for item in range (len(n)):
  n[item] = n[item]+'Z'*(max_len-len(n[item]))

y_train = n
#=================  x_test ==================
n = x_test
j=-1

for i in x_test:
  j=j+1
  if(len(i)>max_len):
    n = np.delete(n, j)
    j=j-1

for item in range (len(n)):
  n[item] = n[item]+'Z'*(max_len-len(n[item]))

x_test = n
#=================  y_test ==================
n = y_test
j=-1

for i in y_test:
  j=j+1
  if(len(i)>max_len):
    n = np.delete(n, j)
    j=j-1

for item in range (len(n)):
  n[item] = n[item]+'Z'*(max_len-len(n[item]))

y_test = n

#============= selected data withing range===========


max_len = max([len(i) for i in x_train])
#print(max_len)

max_len = max([len(i) for i in y_test])
#print(max_len)

print('starting the preprocessing\n')
start_time = time.time()

#==============   Properties Encoded start  ============================

# ==========x_train conversion start====
s = list(x_train)

k = []

for i in range(len(s)):
  t=[]
  for item in range(len(s[i])):
    t.append(prop[ord(s[i][item])-64])
  k.append(t)


x_train = np.array(k)

#=========== x_train conversion end ====

#=========== x_test conversion start====
s = list(x_test)

k = []

for i in range(len(s)):
  t=[]
  for item in range(len(s[i])):
    t.append(prop[ord(s[i][item])-64])
  k.append(t)


x_test = np.array(k)

#============= x_test conversion end====


#==============   Properties Encoded end  ============================



#==============   ONE_HOT   ===================================================

#======= y_train start========
#y_train = y_train[0:3]

alphabet = 'CEHXZ'

char_to_int = dict((c, i) for i, c in enumerate(alphabet))
int_to_char = dict((i, c) for i, c in enumerate(alphabet))

k = []

for i in range(len(y_train)):
  integer_encoded = [char_to_int[char] for char in y_train[i]]

  onehot_encoded=list()
  for value in integer_encoded:
	  letter = [0 for _ in range(len(alphabet))]
	  letter[value] = 1
	  onehot_encoded.append(letter)

  k.append(onehot_encoded)

y_train = np.array(k)
#display(y_train)

#======= y_train end========

#======= y_test start========
#y_train = y_train[0:3]

alphabet = 'CEHXZ'

char_to_int = dict((c, i) for i, c in enumerate(alphabet))
int_to_char = dict((i, c) for i, c in enumerate(alphabet))

k = []

for i in range(len(y_test)):
  integer_encoded = [char_to_int[char] for char in y_test[i]]

  onehot_encoded=list()
  for value in integer_encoded:
	  letter = [0 for _ in range(len(alphabet))]
	  letter[value] = 1
	  onehot_encoded.append(letter)

  k.append(onehot_encoded)

y_test = np.array(k)
#display(y_train)

#======= y_test end========

#==============   ONE_HOT   finish ============================

print('ending the preprocessing\n')
finish_time=time.time()
print ('Time taken to pre-process: ',round(finish_time - start_time,2),' seconds')

#==============   ONE_HOT_INVERSION   =========================================

#for i in range(len(y_train[0])):
#  inverted = int_to_char[argmax(y_train[0][i])]
#  print(inverted)

#================ it's time to learn============================
debug_me()


model=Sequential()

model.add(LSTM((5),batch_input_shape=(None,400,11),return_sequences=True,activation='softmax'))

model.compile(loss='categorical_crossentropy',optimizer='adam',metrics=['accuracy'])

model.summary()

print(model.input_shape)
print(model.output_shape)

history=model.fit(x_train,y_train,epochs=5,validation_data=(x_test,y_test))