Vorpal: A novel RNA virus feature-extraction algorithm demonstrated through interpretable genotype-to-phenotype linear models

Phillip Davis; John Bagnoli; David Yarmosh; Alan Shteyman; Lance Presser; Sharon Altmann; Shelton Bradrick; Joseph A. Russell

In the analysis of genomic sequence data, so-called "alignment free" approaches are often 12 selected for their relative speed compared to alignment-based approaches, especially in the 13 application of distance comparisons and taxonomic classification 1,2,3,4 . These methods are 14 typically reliant on excising K-length substrings of the input sequence, called K-mers 5 . In 15 the context of machine learning, K-mer based feature vectors have been used in 16 applications ranging from amplicon sequencing classification to predictive modeling for 17 antimicrobial resistance genes 6,7,8 . This can be seen as an analogy of the "bag-of-words" 18 model successfully employed in natural language processing and computer vision for 19 document and image classification 9, 10 . Feature extraction techniques from natural language 20 processing have previously been analogized to genomics data 11 ; however, the "bag-of-21