Machine learning detection of SARS-CoV-2 high-risk variants
Li, L.; Li, C.; Li, N.; Zou, D.; Zhao, W.; Xue, Y.; Zhang, Z.; Bao, Y.; Song, S.
Show abstract
The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has evolved many high-risk variants, resulting in repeated COVID-19 waves of pandemic during the past years. Therefore, accurate early-warning of high-risk variants is vital for epidemic prevention and control. Here we construct a machine learning model to predict high-risk variants of SARS-CoV-2 by LightGBM algorithm based on several important haplotype network features. As demonstrated on a series of different retrospective testing datasets, our model achieves accurate prediction of all variants of concern (VOC) and most variants of interest (AUC=0.96). Prediction based on the latest sequences shows that the newly emerging lineage BA.5 has the highest risk score and spreads rapidly to become a major epidemic lineage in multiple countries, suggesting that BA.5 bears great potential to be a VOC. In sum, our machine learning model is capable to early predict high-risk variants soon after their emergence, thus greatly improving public health preparedness against the evolving virus.
Matching journals
The top 2 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Resolving complex structures at oncovirus integration loci with conjugate graph 94%
- LSTM-PHV: Prediction of human-virus protein-protein interactions by LSTM with word2vec 94%
- CoRegNet: Unraveling Gene Co-regulation Networks from Public RNA-Seq Repositories Using a Beta-Binomial Statistical Model 94%
Similar papers in this journal
- GCNCDA: A New Method for Predicting CircRNA-Disease Associations Based on Graph Convolutional Network Algorithm 94%
- Inferring latent temporal progression and regulatory networks from cross-sectional transcriptomic data of cancer samples 94%
- Explainable deep transfer learning model for disease risk prediction using high-dimensional genomic data 93%
Similar papers in this journal
- Both Simulation and Sequencing Data Reveal Multiple SARS-CoV-2 Variants Coinfection in COVID-19 Pandemic 98%
- Real-time monitoring epidemic trends and key mutations in SARS-CoV-2 evolution by an automated tool 93%
- Noncoding RNAs endogenously rule the cancerous regulatory realm while proteins govern the normal 93%