ALT: Another Likelihood Tool

Although there are many phylogenetic reconstruc-tion tools available [1][4], their purpose is in generalto solve specific problems or to apply a limited num-ber of evolution models.We present a new program, ALT — AnotherLikelihood Tool, which implements Felsenstein’smaximum likelihood algorithm [2][3] two importantcharacteristics: it is written in a highly modularfashion and it is extensible.Being modular means that is much easier to makefurther developments in the source code as well todebug the program. As a consequence, improve-ments like optimizations and parallelization can bedone with little effort. Extensibility provides theopportunity of adding new evolution models [5][6]and new heuristics [8] for searching the optimal tree.The same can be said about adapting the programfor dealing with proteins.The original plan was to transform FastDNAml [7]and implement these characteristics, but the fi-nal decision was to develop a new program fromscratch, using new data types and testing alterna-tive algorithms. FastDNAml was used as a refer-ence and testing for correctness. The goal was toobtain the same trees with similar likelihood valuesand hopefully faster.The current version is still approximately 8%slower but the results are very accurate, within 1and 2% when compared with FastDNAml. Theseresults are considered satisfactory as there is stillspace for optimizations. Some parallelization is al-ready implemented by threads and a distributedmemory version is being developed.References [1] J. Felsenstein. Phylip – phylogeny inferencepackage (version 3.2). Cladistics, 5:164–166,1989. [2] J. Felsentein. Maximum likelihood andminimum-steps methods for estimating evolu-tionary trees from data on discrete characters.Syst. Zool., 22:240–249, 1973. [3] J. Felsentein. Evolutionary trees from dna se-quences. J. Mol. Evol., 17:368–376, 1981. [4] Felsenstein J. Phylogeny programs.http://evolution.genetics.washington.edu/phylip/software.html [5] T. H. Jukes and C. R. Cantor. Mam-malian Protein, chapter Evolution of proteinMolecules, pages 21–132. Academic, new yorkedition, 1969. [6] M. Kimura. A simple method for estimatingevolutionary rate of base substitions throughcomparative studies of nucletide sequences. J.Mol. Evol., 16:111–120, 1980. [7] G. J. Olsen, H. Matsuda, Hagstrom R., andR. Oberbeek. A tool for construction of phylo-genetic trees of dna sequences using maximum.Comput. Appl. Biosci., 10:41–48, 1994. [8] D. L. Swofford, G. J. Olsen, P. J. Waddell, andD. M. Hillis. Molecular Systematics, chapterPhylogenetic Inference, pages 407–514. Sin-auer, Sunderland, MA, 2 edition, 1996.