Abstract
Occult hepatitis B infections (OBI) represent a reservoir of undiagnosed and untreated hepatitis B virus (HBV), hence the need to identify mutations that lead to this phenotype. Functionally characterizing these mutations by in vitro studies is time-consuming and expensive. To bridge this gap, in silico approaches, which predict the effect of amino acid (aa) variants on HBV protein function, are necessary. We developed an algorithm for determining the relevance of OBI-associated mutations using in silico approaches. A 3 kb fragment of subgenotypes A1 and D3 from 24 chronic HBV-infected (CHB) and 24 OBI participants was analyzed. To develop and validate the algorithm, the effects of 68 previously characterized occult-associated mutations were determined using three computational tools: PolyPhen2, SNAP2, and PROVEAN. The percentage of deleterious mutations (with impact on protein function) predicted were 52 (76.5%) by PolyPhen2, 55 (80.9%) by SNAP2, and 65 (95.6%) by PROVEAN. At least two tools correctly predicted 59 (86.8%) mutations as deleterious. To identify OBI-associated mutations exclusive to Botswana, study sequences were compared to CHB sequences from GenBank. Of the 43 OBI-associated mutations identified, 26 (60.5%) were predicted by at least two tools to have an impact on protein function. To our knowledge, this is the first study to use in silico approaches to determine the impact of OBI-associated mutations, thereby identifying potential candidates for functional analysis to facilitate mechanistic studies of the OBI phenotype.