ИСТИНА

A common problem with large scale phylogenetic analyses is quality of primary sequence data. Many genomic applications require comparison of multiple phylogenies estimated from different families of orthologous genes in order to infer evolutionary events on a genomic scale. Prediction strength of this type of analysis in many respects will therefore depend upon reliability of individual reconstructions. Disparate substitution rates across regions of homologous sequences and mutational saturation are well known to result in elevated levels of homoplasy in the data and to overshadow available phylogenetic signal. The authors developed a procedure to detect and filter out informational noise from multiple alignment of protein sequence data, thus allowing one to considerably increase accuracy and resolution of phylogenetic analysis. In this work the procedure performance is studied with computer simulations.