top of page

Deep learning moves cancer vaccines toward reality

Researchers at the International Institute of Information Technology Bangalore (IIIT Bangalore – have designed a method using deep learning that can help researchers in immunology take a big leap toward personalized cancer vaccines by an improved understanding of the underlying biological mechanisms of interaction between cancer cells and the body’s immune system.

This research was done by graduate students Gopalakrishnan Venkatesh and Aayush Grover, under the supervision of professors Shrisha Rao and G. Srinivasaraghavan, all of IIIT Bangalore [1]. The research was presented at the 28th Conference on Intelligent Systems in Molecular Biology and was just published in the Bioinformatics journal of the Oxford University Press.

According to the World Health Organization (WHO), cancer is the second leading cause of death worldwide and was responsible for death of an estimated 9.6 million people in 2018 [2]. Research is now focused on personalized cancer vaccines, an approach to help a patient’s own immune system to learn to fight cancer, as a promising weapon in the fight against the disease. The immune system cannot by itself easily distinguish between a healthy and cancerous cell. The way personalized cancer vaccines work is that they externally synthesize a peptide that when passed into the patient helps the immune system identify cancerous cells. This is done by forming a bond between the injected peptide and cancerous cells in the body. Since cancerous cells differ from person to person, such an approach requires analysis to choose the right peptides that can trigger an appropriate immune response.

One of the major steps in the synthesis of personalized cancer vaccines is to computationally predict whether a given peptide will bind with the patient’s Major Histocompatibility Complex (MHC) allele. Peptides and MHC alleles are sequences of amino-acids; peptides are shorter versions of proteins and MHC alleles are proteins essential for the adaptivity of the immune system.

A barrier to the easy development of personalized cancer vaccines is the lack of understanding among the scientific community about how exactly the MHC-peptide binding takes place [4]. Another difficulty is with the need to clinically test different molecules before the vaccine is built, which is resource-intensive task.

This new deep learning model, which the authors call MHCAttnNet, uses Bi-LSTMs [3] to predict the MHC-peptide binding more accurately than existing methods. “Our model is unique in the way that it not only predicts the binding more accurately, but also highlights the subsequences of amino- acids that are likely to be important in order to make a prediction” said Aayush Grover, who is a joint-first author.

MHCAttnNet also uses the attention mechanism, a technique from natural language processing, to highlight the important subsequences from the amino-acid sequences of peptides and MHC alleles that were used by the MHCAttnNet model to make the binding prediction. “If we see how many times a particular subsequence of the allele gets highlighted with a particular amino-acid of peptide, we can learn a lot about the relationship between the peptide and allele subsequences. This would provide insights on how the MHC-peptide binding actually takes place” said Grover. The computational model used in the study has predicted that the number of trigrams of amino-acids of the MHC allele that could be of significance for predicting the binding, corresponding to an amino-acid of a peptide, is plausibly around 3% of the total possible trigrams. This reduced list is enabled by what the authors call "sequence reduction," and will help reduce the work and expense required for clinical trials of vaccines to a large extent.

This work will help researchers develop personalized cancer vaccines by improving the understanding of the MHC-peptide binding mechanism. The higher accuracy of this model will improve the performance of the computational verification step of personalized vaccine synthesis. This, in turn, would improve the likelihood of a personalized cancer vaccine that works on a given patient. Sequence reduction will help focus on a particular few amino acid sequences, which can further facilitate a better understanding of the underlying binding mechanism. Personalized cancer vaccines are still some years away from being available as a mainstream treatment for cancer, and this study offers several directions through sequence reduction that could make it a reality sooner than expected.

The work was supported by an AWS Machine Learning Research Award ( from Amazon. The authors used the AWS Deep Learning machine instances that come pre- installed with popular deep learning frameworks.

“It was a big help that we were able to quickly set up and use high-end machines on Amazon’s AWS cloud for our sophisticated and custom deep learning models, and to easily experiment with new algorithms and approaches,” says Shrisha Rao, professor at IIIT Bangalore, the senior researcher on this study. “It would have cost a fortune to own and operate such hardware outright, and this work is also an illustration of how artificial intelligence and machine learning research using cloud-based solutions can make a mark in different domains including medicine, in a much shorter time and at a fraction of the usual cost.”


Media contacts: Shrisha Rao, E-mail:; Skype: shrisharao

Aayush Grover, E-mail:, Skype: live:85107453010be6a9


[1] - Gopalakrishnan Venkatesh, Aayush Grover, G Srinivasaraghavan, Shrisha Rao (2020). MHCAttnNet: predicting MHC-peptide bindings for MHC alleles classes I and II using an attention-based deep neural model, Bioinformatics, Volume 36, Issue Supplement_1, July 2020, Pages i399– i406,

[2] - WHO Fact Sheet: Cancer (2018). fact-sheets/detail/cancer#:~:text=Key%20facts,%2D%20and%20middle %2Dincome%20countries.

[3] - Schuster, M. and Paliwal, K. (1997). Bidirectional Recurrent Neural Networks. Transactions on Signal Processing, 45(11), 2673–2681, https://

[4] - Rajapakse et al. (2007). Predicting peptides binding to MHC class II molecules using multi-objective evolutionary algorithms. BMC Bioinformatics, 8(1), 459,

1 comment

Recent Posts

See All
bottom of page