sierra-local: A lightweight standalone application for drug resistance prediction

Abstract

Genotypic resistance interpretation systems for the prediction and interpretation of HIV-1 antiretroviral resistance are an important part of the clinical management of HIV-1 infection. Current interpretation systems are generally hosted on remote webservers that enable clinical laboratories to generate resistance predictions easily and quickly from patient HIV-1 sequences encoding the primary targets of modern antiretroviral therapy. However they also potentially compromise a health provider's ethical, professional, and legal obligations to data security, patient information confidentiality, and data provenance. Furthermore, reliance on web-based algorithms makes the clinical management of HIV-1 dependent on a network connection. Here, we describe the development and validation of sierra-local, a local implementation of the Stanford HIVdb genotypic resistance interpretation system, which aims to resolve the ethical, legal, and infrastructure issues associated with remote computing. This open-source package reproduces the HIV-1 resistance scoring by the web-based Stanford HIVdb algorithm with a high degree of concordance (99.990%) and a higher level of performance than previous methods of accessing HIVdb programmatically.