Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MC-NN: An End-to-End Multi-Channel Neural Network Approach for Predicting Influenza A Virus Hosts and Antigenic Types

Yanhua Xu, Dominik Wojtczak

TL;DR

This work addresses predicting Influenza A virus hosts and HA/NA subtypes directly from HA and NA protein sequences using a multi-channel neural network (MC-NN) that fuses two input streams (HA and NA trigrams) to produce three outputs (host, HA subtype, NA subtype). It evaluates three architectures—CNN, BiGRU, and Transformer—via nested cross-validation on pre-2020 data and tests on post-2020 and incomplete sequences, with Transformer generally delivering the strongest performance, including host $F1$ of 83.39% and subtype $F1$ of 99.87% on post-2020 data, and $AP$ of 91.63% with $F1$ of 89.29% on incomplete data. The approach substantially outperforms the BLAST baseline and demonstrates potential for rapid, low-cost surveillance in settings with limited laboratory resources. These results highlight the viability of end-to-end, sequence-based host and subtype prediction and suggest directions for extending the method to cross-species transmission and handling limited data realism.

Abstract

Influenza poses a significant threat to public health, particularly among the elderly, young children, and people with underlying dis-eases. The manifestation of severe conditions, such as pneumonia, highlights the importance of preventing the spread of influenza. An accurate and cost-effective prediction of the host and antigenic sub-types of influenza A viruses is essential to addressing this issue, particularly in resource-constrained regions. In this study, we propose a multi-channel neural network model to predict the host and antigenic subtypes of influenza A viruses from hemagglutinin and neuraminidase protein sequences. Our model was trained on a comprehensive data set of complete protein sequences and evaluated on various test data sets of complete and incomplete sequences. The results demonstrate the potential and practicality of using multi-channel neural networks in predicting the host and antigenic subtypes of influenza A viruses from both full and partial protein sequences.

MC-NN: An End-to-End Multi-Channel Neural Network Approach for Predicting Influenza A Virus Hosts and Antigenic Types

TL;DR

This work addresses predicting Influenza A virus hosts and HA/NA subtypes directly from HA and NA protein sequences using a multi-channel neural network (MC-NN) that fuses two input streams (HA and NA trigrams) to produce three outputs (host, HA subtype, NA subtype). It evaluates three architectures—CNN, BiGRU, and Transformer—via nested cross-validation on pre-2020 data and tests on post-2020 and incomplete sequences, with Transformer generally delivering the strongest performance, including host of 83.39% and subtype of 99.87% on post-2020 data, and of 91.63% with of 89.29% on incomplete data. The approach substantially outperforms the BLAST baseline and demonstrates potential for rapid, low-cost surveillance in settings with limited laboratory resources. These results highlight the viability of end-to-end, sequence-based host and subtype prediction and suggest directions for extending the method to cross-species transmission and handling limited data realism.

Abstract

Influenza poses a significant threat to public health, particularly among the elderly, young children, and people with underlying dis-eases. The manifestation of severe conditions, such as pneumonia, highlights the importance of preventing the spread of influenza. An accurate and cost-effective prediction of the host and antigenic sub-types of influenza A viruses is essential to addressing this issue, particularly in resource-constrained regions. In this study, we propose a multi-channel neural network model to predict the host and antigenic subtypes of influenza A viruses from hemagglutinin and neuraminidase protein sequences. Our model was trained on a comprehensive data set of complete protein sequences and evaluated on various test data sets of complete and incomplete sequences. The results demonstrate the potential and practicality of using multi-channel neural networks in predicting the host and antigenic subtypes of influenza A viruses from both full and partial protein sequences.
Paper Structure (16 sections, 4 equations, 6 figures, 3 tables)

This paper contains 16 sections, 4 equations, 6 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Materials and Methods
  4. Data Preparation
  5. Hemagglutinin and Neuraminidase Protein Sequences
  6. Label Reassignment
  7. Protein Sequence Representation
  8. Neural Network Architectures
  9. Bidirectional Gated Recurrent Unit
  10. Transformer
  11. Convolutional Neural Network
  12. Implementation and Evaluation Methods
  13. Results
  14. Overall Performance
  15. Performance on Single Sequence Input
  16. ...and 1 more sections

Figures (6)

  • Figure 1: Data distribution (hosts)
  • Figure 2: Data distribution (subtypes)
  • Figure 3: The multi-channel neural network architecture: positional encoding is only employed along with Transformer.
  • Figure 4: Comparison of Overall Performance Between Models (Hosts): the baseline results with BLAST are framed by the black solid line.
  • Figure 5: Comparison of Overall Performance Between Models (HA subtypes): the baseline results with BLAST are framed by the black solid line.
  • ...and 1 more figures