Impact of loci nature on estimating recombination and mutation rates in chlamydia trachomatis

Abstract

The knowledge of the frequency and relative weight of mutation and recombination events in evolution is essential for understanding how microorganisms reach fitted phenotypes. Traditionally, these evolutionary parameters have been inferred by using data from multilocus sequence typing (MLST), which is known to have yielded conflicting results. In the near future, these estimations will certainly be performed by computational analyses of full-genome sequences. However, it is not known whether this approach will yield accurate results as bacterial genomes exhibit heterogeneous representation of loci categories, and it is not clear how loci nature impacts such estimations. Therefore, we assessed how mutation and recombination inferences are shaped by loci with different genetic features, using the bacterium Chlamydia trachomatis as the study model. We found that loci assigning a high number of alleles and positively selected genes yielded nonconvergent estimates and incongruent phylogenies and thus are more prone to confound algorithms. Unexpectedly, for the model under evaluation, housekeeping genes and noncoding regions shaped estimations in a similar manner, which points to a nonrandom role of the latter in C. trachomatis evolution. Although the present results relate to a specific bacterium, we speculate that microbe-specific genomic architectures (such as coding capacity, polymorphism dispersion, and fraction of positively selected loci) may differentially buffer the effect of the confounding factors when estimating recombination and mutation rates and, thus, influence the accuracy of using full-genome sequences for such purpose. This putative bias associated with in silico inferences should be taken into account when discussing the results obtained by the analyses of full-genome sequences, in which the “one size fits all” approach may not be applicable.