Chemical encoding examples

Example 1

Example of how to encode a chemical compound as input for a neural network.

The chemical structure is in SMILES format.
Use rdkit to calculate the fingerprint and map into a binary vector.



In [154]:

    
from rdkit import Chem
from rdkit.Chem import AllChem
from rdkit.Chem.Draw import IPythonConsole
from IPython.display import SVG
import pandas as pd



In [155]:

    
import csv
chems = []
with open('data/solubility/delaney.csv') as f:
    cv = csv.DictReader(f)
    for row in cv:
        chems.append( Chem.MolFromSmiles( row['SMILES'] ) )



In [156]:

    
chems[18]









    Out[156]:



In [157]:

    
minPath = 1
maxPath = 5
fpSize = 1024
fp = AllChem.RDKFingerprint(chems[18], minPath=1, maxPath=maxPath, fpSize=fpSize)
px = [int(x) for x in list(fp.ToBitString())]
px[0:20]









    Out[157]:





[0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]

We use synbioTools to map the chemical into a tensor of shape fingerprintSize $\times$ depth. The resulting matrix can be visualized using matplotlib.



In [158]:

    
from synbioTools import tensorChem
depth = 12
fpSize = 20
tc = tensorChem( chems, 20, 4)
tc[0]









    Out[158]:





array([[ 0.,  1.,  0.,  0.],
       [ 0.,  0.,  0.,  0.],
       [ 1.,  0.,  0.,  1.],
       [ 0.,  0.,  1.,  1.],
       [ 0.,  1.,  0.,  0.],
       [ 0.,  0.,  1.,  0.],
       [ 0.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  1.],
       [ 0.,  1.,  0.,  0.],
       [ 0.,  0.,  0.,  0.],
       [ 0.,  0.,  1.,  0.],
       [ 0.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  1.],
       [ 1.,  0.,  1.,  0.],
       [ 1.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  0.],
       [ 0.,  0.,  1.,  0.],
       [ 0.,  1.,  0.,  0.]])



In [159]:

    
%matplotlib inline
import matplotlib.pyplot as plt
plt.imshow(tc[0,:,:])
plt.set_cmap('hot')
plt.xlabel('depth')
plt.ylabel('fingerprint')









    Out[159]:





Text(0,0.5,'fingerprint')

Example 2

Example of encoding an amino-acid sequence into a tensor. We use one-hot encoding for the amino acid and select desired depth for incuding neighboring positions.



In [160]:

    
from Bio import SeqIO
record = list( SeqIO.parse("data/thermostability/l.txt", "fasta") )
seqs = [str(record[i].seq) for i in range(0, len(record)) ]
seqs[0]









    Out[160]:





'MRIKNSGILLLAAILLFSCDKKRVFDEYKSVGSAWHKDSVVTFDLPVLDSTKKYNLFVNLRDNNNYPFNNLFLIVAIETPSGFTKVDTLEYQMANPDGTLMGNGFTDIKESKLYYKEDVKFKGKYKVHIKQAVRESGKIPGVEALEGITDVGFRIEQKD'



In [161]:

    
from synbioTools import tensorSeq
MAX_SEQ_LENGTH = 10
DEPTH = 5
ts, tss = tensorSeq(seqs, MAX_SEQ_LENGTH, DEPTH)



In [162]:

    
%matplotlib inline
import matplotlib.pyplot as plt
plt.imshow(ts[2])
plt.set_cmap('hot')
plt.xlabel('1-hot kmer')
plt.ylabel('position')









    Out[162]:





Text(0,0.5,'position')



In [163]:

    
np.array( aaindex( seqs[0][0:(MAX_SEQ_LENGTH+DEPTH)])









    



  File "<ipython-input-163-8788d97c9cbe>", line 1
    np.array( aaindex( seqs[0][0:(MAX_SEQ_LENGTH+DEPTH)])
                                                         ^
SyntaxError: unexpected EOF while parsing

1-hot encoding for the first amino acids of sequence 0



In [164]:

    
pd.DataFrame( 
    np.vstack( [np.arange(0,MAX_SEQ_LENGTH+DEPTH),
                np.array( aaindex( seqs[0][0:(MAX_SEQ_LENGTH+DEPTH)]) )]
        ), index = ['pos','seq']
)

The sequence window is reversed so that any zero-padding appears at the beginning of the sequence. In the following example we print the sliding window of depth 5 for positions 9,8,7,6:



In [165]:

    
ix = np.arange(0,DEPTH*20, 20)
for a in range(0,4):
    print( np.where(ts[0,a] == 1) - ix )









    



[[9 9 0 0 7]]
[[9 9 9 0 0]]
[[7 9 9 9 0]]
[[5 7 9 9 9]]



In [ ]: