Parsimony and computers

Furthermore, there can be identified two opposing trends in model development. One is a trend toward more detailed models with higher fidelity to the real system, driven by the availability of highly resolved environmental data, increases in computer power, and progress in atmospheric and earth sciences. The other trend is toward models that are tailor-made to specific scientific questions or decision-making problems, driven by the philosophy of parsimony and the increase in the need for scientific results as a basis for decision-making in modem society. 

In view of such developments, it is not surprising that there have been several attempts in 1985-1989 to reconsider the evolutionary relationships. The approach in many studies has been to construct parsimony trees using methods and computer programs based essentially on the maximum parsimony methods of Farris (1970, 1972) or Fitch and Mar-... 

Where appropriate, models were developed with the joint goals of providing a parsimonious explanation of the data while maintaining maximum predictive power. Initial model screening and computation of the r statistic were handled using SAS/GLM. However, for many of the analyses the data contained observations on the same unit measured repeatedly over time. This created a violation of the assumptions for standard statistical techniques and the need to use methodologies for handling models with repeated measures. The tool used in this case was the SAS/MIXED procedure (Littell et al. 1996). 

Felsenstein, J. (1996). Inferring phylogenies from protein sequences by parsimony, distance and Likelihood methods. In Computer Methods for Macromolecular Sequence Analysis (Doolittle, R. F., ed.), Methods Enzymol. 266, 418-427. 

A one standard error rule is described in Hastie et al. (Hastie et al. 2001). It is assumed that several values for the measure of the prediction error at each considered model complexity are available (this can be achieved, e.g., by CV or by bootstrap, Sections 4.2.5 and 4.2.6). Mean and standard error (standard deviation of the means, s) for each model complexity are computed, and the most parsimonious model whose mean prediction error is no more than one standard error above the minimum mean prediction error is chosen. Figure 4.4 (right) illustrates this procedure. The points are the mean prediction errors and the arrows indicate mean plus/minus one standard error. 

To summarize, then, we require a mechanism which (1) describes reaction rate phenomena accurately over a specified range of concentrations, (2) is a parsimonious representation of the actual atmospheric chemistry, in the interest of minimizing computation time, and (3) can be written for a general hydrocarbon species, with the inclusion of variable stoichiometric coefflcients to permit simulation of the behavior of the complex hydrocarbon mixture that actually exists in the atmosphere. Thus, we seek a mechanism which incorporates a balance between accuracy of prediction and ease of computation. 

Distance matrix methods simply count the number of differences between two sequences. This number is referred to as the evolutionary distance, and its exact size depends on the evolutionary model used. The actual tree is then computed from the matrix of distance values by running a clustering algorithm that starts with the most similar sequences (i.e., those that have the shortest distance between them) or by trying to minimize the total branch length of the tree. The principle of maximum parsimony searches for a tree that requires the smallest number of changes to explain the differences observed among the taxa under study. 

---