An improved computational approach to predict HIV protein functions

Sequencing is generating a huge number of sequences, which can be used with sequence matching to derive their functions. Sequence matching is one of the common ways to learn about functions by searching for homologous sequences. If the similarity between sequences is sufficiently high, then functional annotations can be transferred reliably from homologous sequences to the trial protein. Here, we demonstrate the utility of a new substitution matrix (Jia, Jernigan, Proteins, 2021, 89:671), which accounts for inter-dependent amino acid substitutions.