Tristan Lewis Perm: 8959116 EEMB 102 TA: Natasha Picciani EEMB 102 Literature Assignment 1.J P Huelsenbeck; The robustness of two phylogenetic methods: four-taxon simulations reveal a slight superiority of maximum likelihood over neighbor joining., Molecular Biology and Evolution, Volume 12, Issue 5, 1 July 1995, Pages 843–849 2. https://www.researchgate.net/profile/John_Huelsenbeck/publication/15723231_The_Robustness_of_Two_Phylogenetic_Methods_Four-Taxon_Simulations_Reveal_a_Slight_Superiority_of_Maximum_Likelihood_over_Neighbor_Joining/links/58ff2e8c0f7e9bcf65451dcc/The-Robustness-of-Two-Phylogenetic-Methods-Four-Taxon-Simulations-Reveal-a-Slight-Superiority-of-Maximum-Likelihood-over-Neighbor-Joining.pdf 3. 146 other articles have cited this article. My source is “Google Scholar”. 4. I found the article by going onto “Google Scholar” and searching “Phylogenetic Methods” in the search bar. 5. The part of evolution that is being studied in this experiment is the consistency in which the correct phylogeny can be estimated based off of sufficient data. The two data methods being examined are: maximum likelihood and neighbor joining. •Maximum likelihood: a general statistical method for estimating unknown parameters of a probability model. •Neighbor Joining: a method for re-constructing phylogenetic trees, and computing the lengths of the branches of this tree. 6. The main conclusion of the paper is that for any one simulation performed, one can find cases where neighbor joining performs as well as or better than maximum likelihood if one is willing to accept off-diagonal comparisons. Also, it was concluded that both methods had cases in which they failed with high probability. 7. This research utilized a computer system with two models (A no transition/transversion model and a model with an improper transition/transversion ratio) and performed 180 simulation in order to evaluate what model and method is superior in providing consistent results.8. Yes, the data of the simulations is in support of the conclusions made because it concludes that generally speaking maximum likelihood outperformed neighbor joining. In 29 of the 36 cases in which both assumptions for both models was met, the maximum-likelihood value outperformed the neighbor-joining value which supports the conclusion. 9. The results from this study and how virtually any conclusion about the relative performance of phylogenetic methods can be made given inappropriate comparisons shows the need for a systematic way to test the model assumed by phylogenetic methods. 10. Due to how the performance of the model depends greatly on the amount of evolutionary processes generating the characters, a computational model in which all evolutionary processes would be ideal. Also, to further this experiment I believe that varying degrees of altering assumptions of each model in the simulations should be utilized in order to gather further information about the processes generating the characters tested influences the bias of the
View Full Document