Machine learning-based insights on the structural variations found in the C-terminal domains of differing human papillomavirus E6 and E7 protein models across geography, variants, and risk type

Human papillomavirus (HPV) are nonenveloped DNA viruses belonging to the Papillomaviridae family and, during HPV infection, the first viral genes expressed are the E6 and E7 oncogenes with the C-terminal domain mainly involved with oncogenicity. This highlights the importance of elucidating the structural pattern variations found within the E6C and E7C domains in relation to geographic location, viral variants, and risk type. Interestingly, geography-related structural pattern variations within the E6C and E7C domains and its correlation to both viral variants and risk type were never fully elucidated. Here, we generated HPV E6C and E7C protein models and performed superposition across geography, variant, and risk types. Moreover, we performed machine learning analyses (t-SNE, MDS, classification tree, Pythagorean tree) to correlate structural pattern differences across geography, variant, and risk type. We found that some HPV E6C and E7C protein models have similar structural conformation across geography, viral variant, and risk types. Moreover, we found that protein models of HPV58 Japan, HPV68 China, HPV51 Italy, and HPV42 China share structural patterns that potentially helps distinguish the E6C structure and risk type. Similarly, protein models of HPV68 Brazil, HPV51 Italy, HPV40 China, and HPV18 China share structural patterns that potentially help distinguish the E7C structure and risk type. Additionally, we showed that electrostatic surface potential of the E6C and E7C protein surface changes as structural conformation varies which in-turn may likewise serve as a determining feature for risk type. Overall, our results putatively provide insights into how specific features of HPV E6C and E7C may contribute to oncogenic behavior and may support future efforts in molecular risk stratification or targeted therapeutic design.