ORF1ab codon frequency model predicts host-pathogen relationship in orthocoronavirinae

Abstract

Predicting phenotypic properties of a virus directly from its sequence data is an attractive goal for viral epidemiology. Here, we focus narrowly on the Orthocoronavirinae clade and demonstrate models that are powerfully predictive for a human-pathogen phenotype with 76.74% average precision and 85.96% average recall on the withheld test set groups, using only Orf1ab codon frequencies. We show alternative examples for other viral coding sequences and feature representations that do not perform well and discuss what distinguishes the models that are performant. These models point to a small subset of features, specifically 5 codons, that are critical to the success of the models. We discuss and contextualize how this observation may fit within a larger model for the role of translation in virus-host agreement.

Keywords: bioinformactics; feature selection; genotype-to-phenotype; machine learning; viruses.

FIGURE 1

Average performance metrics across each of the one hundred test set splits for each combination of viral CDS and feature representation. Codon frequency model is top performer, with boosts in average performance across each metric over RSCU. Error bars represent 95% confidence interval.

FIGURE 2

Number of Non-Zero (NNZ) coefficients for the top 15 codons in the codon frequency model across all one hundred models fit on Orf1ab for codon frequency and RSCU features. The Trp^TGG codon is used in 96 of 100 codon frequency models but is not available as a feature in the RSCU models.