def score_peptide(peptide, gamma=5):
peptide_feature = protein_features(peptide, lag_range=range(1, 5)).reshape(
(1, -1))
peptide_feature[:] = normalizer.transform(peptide_feature)
return kernel_embedding_score(peptide_feature, gamma)
import pandas as pd
import numpy as np
from Bio import SeqIO
from math import exp
print('Loading the sequences...')
peptides = []
for seq in SeqIO.parse('Data/anti_microbial_peptide.fasta', 'fasta'):
peptides.append(seq)
print('Loading the features...')
if False:
features_reference = np.vstack(
[protein_features(pep, lag_range=range(1, 5)) for pep in peptides])
pd.DataFrame(features_reference).to_csv(
'Data/anti_microbial_peptide_features.csv', index=False)
features_reference = pd.read_csv('Data/anti_microbial_peptide_features.csv')
normalizer = Normalizer()
features_reference = normalizer.fit_transform(features_reference.values)
def kernel_embedding_score(peptide_feature, gamma=0.01):
"""
Calculates the score for kernel embedding
"""
score = np.exp(-gamma * np.sum(
(peptide_feature - features_reference)**2, 1)).mean()