Refinement of phylogenetic signal in multiple sequence alignment: Results of simulation study

Motivation: Disparate substitution rates within the different regions of homologous sequences and mutational saturation are well known to cause misalignment of sequences and to hamper accurate tree reconstruction. Therefore, there is a need in tools detecting and filtering out informational noise from the multiple alignment of sequence data; the tools will help to increase accuracy and resolution of phylogenetic analyses. Results: We propose such a tool and tested its ability to improve the quality phylogenetic trees both on the biological COG data, and on the artificial data, where the ideal tree was known a priory. The key operation of the filtering is a removal of noisy columns. It was shown that the tool permits to reconstruct a tree closer to the "true" tree than is the tree reconstructed with data without removal. Procedure can be applied as a tool to pre-process multiple alignments and enhance phylogenetic inference.