Peptide based vaccine generation system with dual projection generative adversarial networks
A method generates new binding peptides to Major Histocompatibility Complex (MHC) proteins. The method includes training a Generative Adversarial Network (GAN) having a generator and a discriminator only on a set of binding peptide sequences given training data comprising the set of binding peptide sequences and a set of non-binding peptide sequences. A GAN training objective includes the discriminator being iteratively updated to distinguish generated peptide sequences from sampled binding peptide sequences as fake or real and the generator being updated to fool the discriminator. The GAN training objective is optimized while learning two projection vectors for a binding class with two cross-entropy losses. A first loss discriminates binding peptide sequences in the training data from non-binding peptide sequences in the training data. A second loss discriminates generated binding peptide sequences from non-binding peptide sequences in the training data. The method enables supporting decision-making related to new binding peptide generation.
1 . A computer-implemented method for generating new binding peptides to Major Histocompatibility Complex (MHC) proteins, comprising:
training, by a processor device, a Generative Adversarial Network (GAN) having a generator and a discriminator only on a set of binding peptide sequences given training data comprising the set of binding peptide sequences and a set of non-binding peptide sequences, wherein a GAN training objective comprises the discriminator being iteratively updated to distinguish generated peptide sequences from sampled binding peptide sequences as fake or real and the generator being iteratively updated to fool the discriminator;
generating new peptide sequences with user-specified binding properties to create a vaccine based on the trained GAN; and
administering the vaccine,
wherein said training comprises learning two projection vectors for a binding class by optimizing a GAN training objective that combines two cross-entropy losses, a first of the two cross-entropy losses discriminating binding peptide sequences in the training data from non-binding peptide sequences in the training data, and a second of the two cross-entropy losses discriminating generated binding peptide sequences from non-binding peptide sequences in the training data.
2 . The computer-implemented method of claim 1 , wherein the GAN is a Wasserstein GAN.
3 . The computer-implemented method of claim 1 , wherein said training further comprises updating the generator with tempering Softmax units to minimize the two cross-entropy losses.
4 . The computer-implemented method of claim 3 , wherein the tempering Softmax units are employed with entropy regularization for implicit temperature control of the tempering Softmax units.
5 . The computer-implemented method of claim 1 , wherein the generator is a deep neural network, comprising a convolutional layer for receiving the set of binding peptide sequences.
6 . The computer-implemented method of claim 1 , wherein the GAN operates on sampled latent code vectors from a multivariate Gaussian distribution obtained from the training data.
7 . The computer-implemented method of claim 1 , wherein the two cross-entropy losses are implemented by dual classifiers.
8 . The computer-implemented method of claim 1 , wherein the new peptide sequences are output from the generator as Softmax output units, and wherein the generator comprises a fully-connected layer for receiving an input random noise vector and another fully-connected layer for outputting the Softmax output units.
9 . A computer program product for generating new binding peptides to Major Histocompatibility Complex (MHC) proteins, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform a method comprising:
training, by a processor device of the computer, a Generative Adversarial Network (GAN) having a generator and a discriminator only on a set of binding peptide sequences given training data comprising the set of binding peptide sequences and a set of non-binding peptide sequences, wherein a GAN training objective comprises the discriminator being iteratively updated to distinguish generated peptide sequences from sampled positive peptide sequences as fake or real and the generator being iteratively updated to fool the discriminator;
generating new peptide sequences with user-specified binding properties to create a vaccine based on the trained GNA; and
administering the vaccine,
wherein said training comprises learning two projection vectors for a binding class by optimizing a GAN training objective that combines two cross-entropy losses, a first of the two cross-entropy losses discriminating binding peptide sequences in the training data from non-binding peptide sequences in the training data, and a second of the two cross-entropy losses discriminating generated binding peptide sequences from non-binding peptide sequences in the training data.
10 . The computer program product of claim 9 , wherein the GAN is a Wasserstein GAN.
11 . The computer program product of claim 9 , wherein said training further comprises updating the generator with tempering Softmax units to minimize the two cross-entropy losses.
12 . The computer program product of claim 11 , wherein the tempering Softmax units are employed with entropy regularization for implicit temperature control of the tempering Softmax units.
13 . The computer program product of claim 9 , wherein the generator is a deep neural network, comprising a convolutional layer for receiving the set of binding peptide sequences.
14 . The computer program product of claim 9 , wherein the GAN operates on sampled latent code vectors from a multivariate Gaussian distribution obtained from the training data.
15 . The computer program product of claim 9 , wherein the two cross-entropy losses are implemented by dual classifiers.
16 . The computer program product of claim 9 , wherein the new peptide sequences are output from the generator as Softmax output units, and wherein the generator comprises a fully-connected layer for receiving an input random noise vector and another fully-connected layer for outputting the Softmax output units.
17 . A computer processing system for generating new binding peptides to Major Histocompatibility Complex (MHC) proteins, comprising:
a memory device for storing program code; and
a processor device, operatively coupled to the memory device, for running the program code to:
train a Generative Adversarial Network (GAN) having a generator and a discriminator only on a set of binding peptide sequences given training data comprising the set of binding peptide sequences and a set of non-binding peptide sequences, wherein a GAN training objective comprises the discriminator being iteratively updated to distinguish generated peptide sequences from sampled positive binding peptide sequences as fake or real and the generator being iteratively updated to fool the discriminator;
generate new peptide sequences with user-specified binding properties to create a vaccine based on the trained GAN; and
administer the vaccine,
wherein, during training, the processor device further runs the program code to learn two projection vectors for a binding class by optimizing a GAN training objective that combines two cross-entropy losses, a first of the two cross-entropy losses discriminating binding peptide sequences in the training data from non-binding peptide sequences in the training data, and a second of the two cross-entropy losses discriminating generated binding peptide sequences from non-binding peptide sequences in the training data.
18 . The computer processing system of claim 17 , wherein said processor device further runs the program code to use the generated positive binding peptides as additional positive training data to improve a performance of a classifier discriminating binding peptides from non-binding peptides.
19 . The computer-implemented method of claim 1 , wherein the cross-entropy losses are expressed as:
L
m
i
p
=
-
v
y
p
T
ϕ
(
x
+
)
+
log
∑
y
′
e
v
y
′
v
p
T
ϕ
(
x
+
)
L
m
i
q
=
-
v
y
q
T
ϕ
(
x
-
)
+
log
∑
y
′
e
v
y
′
q
T
ϕ
(
x
-
)
where p and q correspond to conditional distribution or loss function using real/generated binding peptides, the terms v y p and v y q represent embeddings of the real and generated samples, respectively, ϕ(·) is an embedding function, x + ˜P X and x − ˜Q X are real and generated sequences (with P and Q being the respective real and generated distributions), and y is a data label.