Drug Resistance Prediction Using Deep Learning Techniques on HIV-1 Sequence Data
The fast replication rate and lack of repair mechanisms of human immunodeficiency virus (HIV) contribute to its high mutation frequency, with some mutations resulting in the evolution of resistance to antiretroviral therapies (ART). As such, studying HIV drug resistance allows for real-time evaluati...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2020-05-01
|
Series: | Viruses |
Subjects: | |
Online Access: | https://www.mdpi.com/1999-4915/12/5/560 |
_version_ | 1797567584883703808 |
---|---|
author | Margaret C. Steiner Keylie M. Gibson Keith A. Crandall |
author_facet | Margaret C. Steiner Keylie M. Gibson Keith A. Crandall |
author_sort | Margaret C. Steiner |
collection | DOAJ |
description | The fast replication rate and lack of repair mechanisms of human immunodeficiency virus (HIV) contribute to its high mutation frequency, with some mutations resulting in the evolution of resistance to antiretroviral therapies (ART). As such, studying HIV drug resistance allows for real-time evaluation of evolutionary mechanisms. Characterizing the biological process of drug resistance is also critically important for sustained effectiveness of ART. Investigating the link between “black box” deep learning methods applied to this problem and evolutionary principles governing drug resistance has been overlooked to date. Here, we utilized publicly available HIV-1 sequence data and drug resistance assay results for 18 ART drugs to evaluate the performance of three architectures (multilayer perceptron, bidirectional recurrent neural network, and convolutional neural network) for drug resistance prediction, jointly with biological analysis. We identified convolutional neural networks as the best performing architecture and displayed a correspondence between the importance of biologically relevant features in the classifier and overall performance. Our results suggest that the high classification performance of deep learning models is indeed dependent on drug resistance mutations (DRMs). These models heavily weighted several features that are not known DRM locations, indicating the utility of model interpretability to address causal relationships in viral genotype-phenotype data. |
first_indexed | 2024-03-10T19:45:00Z |
format | Article |
id | doaj.art-a07a148abcb14989b7fa14743006b568 |
institution | Directory Open Access Journal |
issn | 1999-4915 |
language | English |
last_indexed | 2024-03-10T19:45:00Z |
publishDate | 2020-05-01 |
publisher | MDPI AG |
record_format | Article |
series | Viruses |
spelling | doaj.art-a07a148abcb14989b7fa14743006b5682023-11-20T00:55:25ZengMDPI AGViruses1999-49152020-05-0112556010.3390/v12050560Drug Resistance Prediction Using Deep Learning Techniques on HIV-1 Sequence DataMargaret C. Steiner0Keylie M. Gibson1Keith A. Crandall2Computational Biology Institute, Milken Institute School of Public Health, The George Washington University, Washington, DC 20052, USAComputational Biology Institute, Milken Institute School of Public Health, The George Washington University, Washington, DC 20052, USAComputational Biology Institute, Milken Institute School of Public Health, The George Washington University, Washington, DC 20052, USAThe fast replication rate and lack of repair mechanisms of human immunodeficiency virus (HIV) contribute to its high mutation frequency, with some mutations resulting in the evolution of resistance to antiretroviral therapies (ART). As such, studying HIV drug resistance allows for real-time evaluation of evolutionary mechanisms. Characterizing the biological process of drug resistance is also critically important for sustained effectiveness of ART. Investigating the link between “black box” deep learning methods applied to this problem and evolutionary principles governing drug resistance has been overlooked to date. Here, we utilized publicly available HIV-1 sequence data and drug resistance assay results for 18 ART drugs to evaluate the performance of three architectures (multilayer perceptron, bidirectional recurrent neural network, and convolutional neural network) for drug resistance prediction, jointly with biological analysis. We identified convolutional neural networks as the best performing architecture and displayed a correspondence between the importance of biologically relevant features in the classifier and overall performance. Our results suggest that the high classification performance of deep learning models is indeed dependent on drug resistance mutations (DRMs). These models heavily weighted several features that are not known DRM locations, indicating the utility of model interpretability to address causal relationships in viral genotype-phenotype data.https://www.mdpi.com/1999-4915/12/5/560HIVantiretroviral therapyHIV drug resistancemachine learningdeep learningneural networks |
spellingShingle | Margaret C. Steiner Keylie M. Gibson Keith A. Crandall Drug Resistance Prediction Using Deep Learning Techniques on HIV-1 Sequence Data Viruses HIV antiretroviral therapy HIV drug resistance machine learning deep learning neural networks |
title | Drug Resistance Prediction Using Deep Learning Techniques on HIV-1 Sequence Data |
title_full | Drug Resistance Prediction Using Deep Learning Techniques on HIV-1 Sequence Data |
title_fullStr | Drug Resistance Prediction Using Deep Learning Techniques on HIV-1 Sequence Data |
title_full_unstemmed | Drug Resistance Prediction Using Deep Learning Techniques on HIV-1 Sequence Data |
title_short | Drug Resistance Prediction Using Deep Learning Techniques on HIV-1 Sequence Data |
title_sort | drug resistance prediction using deep learning techniques on hiv 1 sequence data |
topic | HIV antiretroviral therapy HIV drug resistance machine learning deep learning neural networks |
url | https://www.mdpi.com/1999-4915/12/5/560 |
work_keys_str_mv | AT margaretcsteiner drugresistancepredictionusingdeeplearningtechniquesonhiv1sequencedata AT keyliemgibson drugresistancepredictionusingdeeplearningtechniquesonhiv1sequencedata AT keithacrandall drugresistancepredictionusingdeeplearningtechniquesonhiv1sequencedata |