ИСТИНА

Motivation: Disparate substitution rates across regions of homologous sequences and mutational saturation are well known to cause misalignment of sequences and to hamper accurate tree reconstruction. In the meantime, tools are missing to detect and filter out informational noise from the multiple alignment of sequence data in order to increase accuracy and resolution of phylogenetic analyses. Results: A dedicated procedure was developed by the authors and tested initially with biological COG data. In this study it is verified in simulation experiments, in which the ideal tree is known a priory. Removing noisy columns permits to reconstruct a tree closer to the known tree than is the tree reconstructed with data without removal. Procedure can be applied as a tool to process multiple alignments and enhance phylogenetic inference.