A/The Brief History of Three-Taxon Analysis

Recent claims by its advocates notwithstanding, three-taxon analysis (3ta) provides no method for recognizing reversals or for applying them as apomorphies. Accordingly, 3ta could be used as a phylogenetic method only under an assumption of irreversibility. Being a method for calculating trees from character data, 3ta is not connected to any particular rule (“interpretation”) for selecting resolutions of consensus trees considered as abstract diagrams. 3ta cannot be justified simply by invoking a general minimization principle such as Occam's Razor, since that would cover almost any method. Some more specific basis is needed, and consideration of proposed bases for 3ta shows that none is even remotely adequate.     

The Parsimony Ratchet, a New Method for Rapid Parsimony Analysis

The Parsimony Ratchet ¹ 1 This method, the Parsimony Ratchet, was originally presented at the Numerical Cladistics Symposium at the American Museum of Natural History, New York, in May 1998 (see Horovitz, 1999) and at the Meeting of the Willi Hennig Society (Hennig XVII) in September 1998 in São Paulo, Brazil.
is presented as a new method for analysis of large data sets. The method can be easily implemented with existing phylogenetic software by generating batch command files. Such an approach has been implemented in the programs DADA (Nixon, 1998) and Winclada (Nixon, 1999). The Parsimony Ratchet has also been implemented in the most recent versions of NONA (Goloboff, 1998). These implementations of the ratchet use the following steps: (1) Generate a starting tree (e.g., a “Wagner” tree followed by some level of branch swapping or not). (2) Randomly select a subset of characters, each of which is given additional weight (e.g., add 1 to the weight of each selected character). (3) Perform branch swapping (e.g., “branch-breaking” or TBR) on the current tree using the reweighted matrix, keeping only one (or few) tree. (4) Set all weights for the characters to the “original” weights (typically, equal weights). (5) Perform branch swapping (e.g., branch-breaking or TBR) on the current tree (from step 3) holding one (or few) tree. (6) Return to step 2. Steps 2–6 are considered to be one iteration, and typically, 50–200 or more iterations are performed. The number of characters to be sampled for reweighting in step 2 is determined by the user; I have found that between 5 and 25% of the characters provide good results in most cases. The performance of the ratchet for large data sets is outstanding, and the results of analyses of the 500 taxon seed plant rbcL data set (Chase et al., 1993) are presented here. A separate analysis of a three-gene data set for 567 taxa will be presented elsewhere (Soltis et al., in preparation) demonstrating the same extraordinary power. With the 500-taxon data set, shortest trees are typically found within 22 h (four runs of 200 iterations) on a 200-MHz Pentium Pro. These analyses indicate efficiency increases of 20×–80× over “traditional methods” such as varying taxon order randomly and holding few trees, followed by more complete analyses of the best trees found, and thousands of times faster than nonstrategic searches with PAUP. Because the ratchet samples many tree islands with fewer trees from each island, it provides much more accurate estimates of the “true” consensus than collecting many trees from few islands. With the ratchet, Goloboff's NONA, and existing computer hardware, data sets that were previously intractable or required months or years of analysis with PAUP* can now be adequately analyzed in a few hours or days.     

A Parsimony Approach to Biological Pathway Reconstruction/Inference for Genomes and Metagenomes

A common biological pathway reconstruction approach—as implemented by many automatic biological pathway services (such as the KAAS and RAST servers) and the functional annotation of metagenomic sequences—starts with the identification of protein functions or families (e.g., KO families for the KEGG database and the FIG families for the SEED database) in the query sequences, followed by a direct mapping of the identified protein families onto pathways. Given a predicted patchwork of individual biochemical steps, some metric must be applied in deciding what pathways actually exist in the genome or metagenome represented by the sequences. Commonly, and straightforwardly, a complete biological pathway can be identified in a dataset if at least one of the steps associated with the pathway is found. We report, however, that this naïve mapping approach leads to an inflated estimate of biological pathways, and thus overestimates the functional diversity of the sample from which the DNA sequences are derived. We developed a parsimony approach, called MinPath (Minimal set of Pathways), for biological pathway reconstructions using protein family predictions, which yields a more conservative, yet more faithful, estimation of the biological pathways for a query dataset. MinPath identified far fewer pathways for the genomes collected in the KEGG database—as compared to the naïve mapping approach—eliminating some obviously spurious pathway annotations. Results from applying MinPath to several metagenomes indicate that the common methods used for metagenome annotation may significantly overestimate the biological pathways encoded by microbial communities.     

On the TTSC–FTSC Formulation of Standard Parsimony

Standard parsimony analysis has recently been described in a “three-taxon-like” way (the three-taxa statements for contiguous series–four-taxa statements for contiguous series, or TTSC–FTSC procedure) in order to clarify the differences between the standard approach and three-taxon analysis. It is shown that the alleged equivalence of standard parsimony analysis and the TTSC–FTSC procedure does not hold. Some minor defects of the procedure can be fixed within the TTSC–FTSC logic, but no solution is available for two basic problems: (1) the elementary three-taxon-like statements of the TTSC–FTSC procedure are highly artificial; and (2) the equivalence with standard parsimony depends on an incomplete correction for nonindependence between these statements. However, these findings do not invalidate the reported superiority of standard parsimony as a method for biological systematics.     

Parsimony via consensus

The parsimony score of a character on a tree equals the number of state changes required to fit that character onto the tree. We show that for unordered, reversible characters this score equals the number of tree rearrangements required to fit the tree onto the character. We discuss implications of this connection for the debate over the use of consensus trees or total evidence and show how it provides a link between incongruence of characters and recombination.     

Using Genes as Characters and a Parsimony Analysis to Explore the Phylogenetic Position of Turtles

The phylogenetic position of turtles within the vertebrate tree of life remains controversial. Conflicting conclusions from different studies are likely a consequence of systematic error in the tree construction process, rather than random error from small amounts of data. Using genomic data, we evaluate the phylogenetic position of turtles with both conventional concatenated data analysis and a “genes as characters” approach. Two datasets were constructed, one with seven species (human, opossum, zebra finch, chicken, green anole, Chinese pond turtle, and western clawed frog) and 4584 orthologous genes, and the second with four additional species (soft-shelled turtle, Nile crocodile, royal python, and tuatara) but only 1638 genes. Our concatenated data analysis strongly supported turtle as the sister-group to archosaurs (the archosaur hypothesis), similar to several recent genomic data based studies using similar methods. When using genes as characters and gene trees as character-state trees with equal weighting for each gene, however, our parsimony analysis suggested that turtles are possibly sister-group to diapsids, archosaurs, or lepidosaurs. None of these resolutions were strongly supported by bootstraps. Furthermore, our incongruence analysis clearly demonstrated that there is a large amount of inconsistency among genes and most of the conflict relates to the placement of turtles. We conclude that the uncertain placement of turtles is a reflection of the true state of nature. Concatenated data analysis of large and heterogeneous datasets likely suffers from systematic error and over-estimates of confidence as a consequence of a large number of characters. Using genes as characters offers an alternative for phylogenomic analysis. It has potential to reduce systematic error, such as data heterogeneity and long-branch attraction, and it can also avoid problems associated with computation time and model selection. Finally, treating genes as characters provides a convenient method for examining gene and genome evolution.     

On the Parametric Approach to Survival Data Analysis I. Radiobiological Applications

Acute radiation injury and postirradiation recovery have been formalized in terms of a Markovian homogeneous process of the random walk with a finite set of states, two absorbing barriers and continuous time. The distribution of time for such a process to reach (for the first time) the upper absorbing barrier was earlier obtained by SAATY (1961) and within the proposed model it coincides with the life span distribution for irradiated animals. The possibilities of finding the maximum likelihood estimates of unknown parameters are investigated by means of simulating experiments performed with a computer assistance. On the basis of simulation results the applicability of the proposed distribution for the purposes of survival data analysis is discussed. Extension of the model to accomodate two (or more) radiation syndromes is presented.     

Analysis of the Distribution of Bootstrap Tree Lengths Using the Maximum Parsimony Method

The bootstrap is an important tool for estimating the confidence interval of monophyletic groups within phylogenies. Although bootstrap analyses are used in most evolutionary studies, there is no clear consensus as how best to interpret bootstrap probability values. To study further the bootstrap method, nine small subunit ribosomal DNA (SSU rDNA) data sets were submitted to bootstrapped maximum parsimony (MP) analyses using unweighted and weighted sequence positions. Analyses of the lengths (i.e., parsimony steps) of the bootstrap trees show that the shape and mean of the bootstrap tree distribution may provide important insights into the evolutionary signal within the sequence data. With complex phylogenies containing nodes defined by short internal branches (multifurcations), the mean of the bootstrap tree distribution may differ by 2 standard deviations from the length of the best tree found from the original data set. Weighting sequence positions significantly increases the bootstrap values at internal nodes. There may, however, be strong bootstrap support for conflicting species groupings among different data sets. This phenomenon appears to result from a correlation between the topology of the tree used to create the weights and the topology of the bootstrap consensus tree inferred from the MP analysis of these weighted data. The analyses also show that characteristics of the bootstrap tree distribution (e.g., skewness) may be used to choose between alternative weighting schemes for phylogenetic analyses.     

Parsimony, simplicity and what actually happened

The role of a parsimony principle is unclear in most methods which have been claimed to be valid for the reconstruction of tionary kinship. There appear to be two reasons for this: first, the role of parsimony is generally uncertain in scientific method; second, the majority of methods proposed transform data and order them, but are not appropriate to the reconstruction of phyto Commitment to a probabilistic model of tionary processes seems to be the essential component which may enable us justifiably to estimate phylo An example is provided which emphasizes the importance of knowledge about the nature of the process before undertaking estimation of the pattern of kinship.     

Parsimony Analysis as a Specific Kind of Homology Estimation and the Implications for Character Weighting

“Remane-Hennigian systematists” still reject parsimony analysis for phylogenetics, because homology or apomorphy analyses are not included. In contrast, “pattern cladists” regard homology as a deductive concept after applying a parsimony test of character congruence. However, as in molecular phylogeny, selection of “good” characters is always done on the basis of ana priorihomology analysis. The distribution criterion of homology—“homologous characters have identical or hierarchical distribution”—is the basis of parsimony analysis. Because this criterion also might fail in cases of genealogical reticulation or concerted homoplasy, character congruence is not a strict test but another probabilistic criterion of homology. A synthetic approach is proposed for phenotypic analysis with application ofa prioricriteria of homology. The resultinga prioriprobabilities of homology serve as criteria for selection and weighting of characters (very low = not selected/poor/mediocre/good/Dollo characters). After application of a parsimony algorithm the final cladogram decides homology estimations.     

Approximate Maximum Parsimony and Ancestral Maximum Likelihood

We explore the maximum parsimony (MP) and ancestral maximum likelihood (AML) criteria in phylogenetic tree reconstruction. Both problems are NP-hard, so we seek approximate solutions. We formulate the two problems as Steiner tree problems under appropriate distances. The gist of our approach is the succinct characterization of Steiner trees for a small number of leaves for the two distances. This enables the use of known Steiner tree approximation algorithms. The approach leads to a 16/9 approximation ratio for AML and asymptotically to a 1.55 approximation ratio for MP.     

A Variational Bayes Approach to the Analysis of Occupancy Models

Detection-nondetection data are often used to investigate species range dynamics using Bayesian occupancy models which rely on the use of Markov chain Monte Carlo (MCMC) methods to sample from the posterior distribution of the parameters of the model. In this article we develop two Variational Bayes (VB) approximations to the posterior distribution of the parameters of a single-season site occupancy model which uses logistic link functions to model the probability of species occurrence at sites and of species detection probabilities. This task is accomplished through the development of iterative algorithms that do not use MCMC methods. Simulations and small practical examples demonstrate the effectiveness of the proposed technique. We specifically show that (under certain circumstances) the variational distributions can provide accurate approximations to the true posterior distributions of the parameters of the model when the number of visits per site (K) are as low as three and that the accuracy of the approximations improves as K increases. We also show that the methodology can be used to obtain the posterior distribution of the predictive distribution of the proportion of sites occupied (PAO).     

A Mathematical Approach to the Analysis of Multiplex DNA Profiles

Multiplex DNA profiles are used extensively for biomedical and forensic purposes. However, while DNA profile data generation is automated, human analysis of those data is not, and the need for speed combined with accuracy demands a computer-automated approach to sample interpretation and quality assessment. In this paper, we describe an integrated mathematical approach to modeling the data and extracting the relevant information, while rejecting noise and sample artifacts. We conclude with examples showing the effectiveness of our algorithms.     

A Global Optimization Approach to Multi-Polarity Sentiment Analysis

Following the rapid development of social media, sentiment analysis has become an important social media mining technique. The performance of automatic sentiment analysis primarily depends on feature selection and sentiment classification. While information gain (IG) and support vector machines (SVM) are two important techniques, few studies have optimized both approaches in sentiment analysis. The effectiveness of applying a global optimization approach to sentiment analysis remains unclear. We propose a global optimization-based sentiment analysis (PSOGO-Senti) approach to improve sentiment analysis with IG for feature selection and SVM as the learning engine. The PSOGO-Senti approach utilizes a particle swarm optimization algorithm to obtain a global optimal combination of feature dimensions and parameters in the SVM. We evaluate the PSOGO-Senti model on two datasets from different fields. The experimental results showed that the PSOGO-Senti model can improve binary and multi-polarity Chinese sentiment analysis. We compared the optimal feature subset selected by PSOGO-Senti with the features in the sentiment dictionary. The results of this comparison indicated that PSOGO-Senti can effectively remove redundant and noisy features and can select a domain-specific feature subset with a higher-explanatory power for a particular sentiment analysis task. The experimental results showed that the PSOGO-Senti approach is effective and robust for sentiment analysis tasks in different domains. By comparing the improvements of two-polarity, three-polarity and five-polarity sentiment analysis results, we found that the five-polarity sentiment analysis delivered the largest improvement. The improvement of the two-polarity sentiment analysis was the smallest. We conclude that the PSOGO-Senti achieves higher improvement for a more complicated sentiment analysis task. We also compared the results of PSOGO-Senti with those of the genetic algorithm (GA) and grid search method. From the results of this comparison, we found that PSOGO-Senti is more suitable for improving a difficult multi-polarity sentiment analysis problem.     

Bayesian Analysis Using a Simple Likelihood Model Outperforms Parsimony for Estimation of Phylogeny from Discrete Morphological Data

Despite the introduction of likelihood-based methods for estimating phylogenetic trees from phenotypic data, parsimony remains the most widely-used optimality criterion for building trees from discrete morphological data. However, it has been known for decades that there are regions of solution space in which parsimony is a poor estimator of tree topology. Numerous software implementations of likelihood-based models for the estimation of phylogeny from discrete morphological data exist, especially for the Mk model of discrete character evolution. Here we explore the efficacy of Bayesian estimation of phylogeny, using the Mk model, under conditions that are commonly encountered in paleontological studies. Using simulated data, we describe the relative performances of parsimony and the Mk model under a range of realistic conditions that include common scenarios of missing data and rate heterogeneity.