Limitations in Computational Analysis of Retrovirus Evolution

Abstract

Retroviruses and retrotransposons with long terminal repeats (a type of transposable elements in eukaryotic genome) are very similar in their structure and life cycle that strongly indicates their common origin. Obviously, one of the structures transformed into others in the process of evolution. However, it is not clear which of the structures appeared earlier. There are two quite convincing scenarios of evolution: a scenario describing the ways of transformation of retrotransposons into retroviruses and the reverse scenario. The Drosophila melanogaster genome provides an excellent opportunity to analyze both possible scenarios for the evolution of retroelements, since it, unlike, for example, the human genome, is filled with diverse families of functionally active retrotransposons, including retrotransposons–retroviruses with infectious properties. The construction of evolutionary models—evolutionary trees—requires alignment of conserved amino acid sequences encoded by both types of retroelements. For phylogenetic trees construction, a variety of algorithms are developed. The most reliable approach is based on the principle of maximum likelihood. However, in the process of computational analysis, we meet with several problems that algorithms for constructing phylogenetic trees usually ignore. On the example of the model object, Drosophila, we consider the main limitations of modeling the evolution of retroelements: a tendency to accumulation of repeats, a horizontal transfer of sequences, and a rate of viral sequence evolution.

This work is supported by RFBR, pr. N 17-04-01250.