The p-Epitope Model of Influenza Antigenic Distance

Applied Physics Graduate Program, Rice University (Houston, TX)

M.E. Bonomo, R.Y. Kim, & M.W. Deem

Doctoral Research, Spring 2016 – Spring 2019

The influenza virus is constantly evolving, necessitating that the vaccine is re-evaluated each flu season.  The vaccine helps the immune system to prepare antibodies against the influenza virus, so it will be most effective if the flu strain used to make the vaccine matches the flu strains that are circulating.

In the human immune system, antibodies recognize the influenza virus by way of five modular sites on the virus’s surface, called epitopes. We are developing a mathematical model, called pEpitope (pronounced PEE-epih-tope), that exploits this premise in order to predict the efficacy of the seasonal flu vaccine.

Download:

• pEpitope Calculator for MATLAB

In the News: 

• November 8, 2018 – Math can improve flu vaccine, Baker Institute experts say
• October 2, 2018 – Rice Expert Alert: Will this year’s flu shot work?
• April 19, 2018 – Study predicts 2018 flu vaccine will have 19 percent efficacy

Publications: 

• M.E. Bonomo, R.Y. Kim, and M.W. Deem, “Modular epitope binding predicts influenza quasispecies dominance and vaccine effectiveness: Application to 2018/19 season,” Vaccine, 37(24): 3154-3158, 2019.
• M.W. Deem, M.E. Bonomo, and K.R.W. Matthews, “Improving the Effectiveness of the Annual Flu Vaccine,” Policy brief no. 10.29.18. Rice University’s Baker Institute for Public Policy, Houston, Texas.
• M.E. Bonomo and M.W. Deem, “Reply to Xu and Ye: On the sufficiency of the pEpitope method,” Clinical Infectious Diseases, 68(2): 347, 2019.
• M.E. Bonomo and M.W. Deem, “Predicting influenza H3N2 vaccine efficacy from evolution of the dominant epitope,” Clinical Infectious Diseases, 67(7): 1129-1131, 2018.

Oral Presentations:

• American Physical Society Annual March Meeting (Boston, MA) 6 March 2019.
• Fall Meeting of the Texas Section of the American Physical Society (Houston, TX) 20 October 2018. (Best Paper Honors)
• 12th Annual q-bio (Quantitative Biology) Conference (Houston, TX) 29 June 2018.
• 3rd Annual Smalley-Curl Institute Transdisciplinary Symposium (Houston, TX) 9 February 2018. (Presenter Award)

---

Purpose

The pEpitope variable measures the antigenic distance between two influenza strains. The Deem Group conceived of pEpitope and illustrated how this method shows high correlation to influenza vaccine efficacy data. The pEpitope Calculator program is designed to facilitate use of this measure of antigenic distance.

Description

The MATLAB file that can be downloaded below will install an app in MATLAB to calculate pEpitope for discrepancies in the hemagglutinin proteins between the vaccine and dominant circulating strains of Influenza Type A H3N2 viruses and to estimate vaccine efficacy.

• Version 1.0 : Vaccine efficacy is predicted based on the model in Bonomo and Deem, 2018 (CID, 67(7): 1129-1131).
• Version 2.0 : Vaccine efficacy is predicted based on the updated model in Bonomo et al., 2019 (Vaccine, 37(24): 3154-3158), and contains an option to use data in test-negative design studies from the Centers for Disease Control and Prevention (CDC) United States Vaccine Effectiveness Network or data in studies over the past decade from national influenza surveillance networks across the northern hemisphere (NH).
• Version 2.1: Includes the option to calculate vaccine effectiveness.

On the main “pEpitope Calculator” tab of the app, the user inputs two amino acid sequences of the hemagglutinin protein for comparison using one-letter IUPAC abbreviations. The program is designed to take input in the form of pasted sequences containing valid amino acid characters. The strains are then aligned using a reference strain: A/Hong Kong/4801/2014 (EPI614406) for A/H3N2 human.

If the program encounters major problems aligning the sequences, then an error message will be displayed requesting that the user try again. The most likely reason for this problem is that the hemagglutinin type of the input strain does not match that of the reference. If the alignment of one of the user’s strains leaves extra characters before or after the reference strain, these characters are discarded. If the alignment has the user’s strain starting after or ending before the reference strain, then these offsets are noted, as they could affect pEpitope. An error message is displayed that states that the sequences are “incomplete” if they contain epitope regions.

The GNU General Public License is found on the “GNL” tab.

Download

The .mlappinstall file is an archive file for sharing the app created using MATLAB. It contains everything necessary to install and run the app: the source code, supporting data, information (such as product dependencies), and the app icon.

In any publications that result from the use of the pEpitope Calculator (Version 2.1), please cite:

M.E. Bonomo, R.Y. Kim, M.W. Deem, “Modular epitope binding predicts influenza quasispecies dominance and vaccine effectiveness: Application to 2018/19 season,” Vaccine, 37(24): 3154-3158, 2019.

Usage

Instructions for installing the pEpitope Calculator app (requires MATLAB R2012b or later):

• Download the pEpitopeCalculator.mlappinstall file to your computer.
• Manual Installation:
  • In MATLAB, run this command: appinfo = matlab.apputil.install(‘pEpitopeCalculator.mlappinstall’)
• Automatic Installation:
  • In MATLAB, under the “Apps” tab, select “Install App.”
  • Select the pEpitopeCalculator.mlappinstall file.

After installing the app, you can paste strain names and sequences containing valid amino acid characters for Influenza A H3N2 viruses and click “Calculate.” The value of pEpitope and the dominant epitope are output, along with the positions and amino acids of mutations in each epitope. The predicted vaccine efficacy and standard error are also output. The “Reset” button will clear the form for subsequent calculations.

You can download a text file of each sequence of interest from GISAID or the NIH Influenza Virus Resource. For example, to copy/paste sequences into the program, do the following:

1. Go to the NIH Influenza Virus Resource.
2. Search for the sequence of the first strain you want to compare. Click its Accession number to view the strain report.
3. Select “FASTA” to view and copy the amino acid sequence.
4. Next, in MATLAB paste this sequence into the appropriate text box. Remove line breaks as necessary.
5. Repeat these steps for the second strain.
6. Click “Calculate” and await results.

Feedback

The pEpitope Calculator is unsupported software. Please send questions and comments to M. E. Bonomo.