A data management framework for the AFTOL project

Cox, C., Kauff, F. & Lutzoni, F.
Duke University, Dept. of Biology, Durham NC-27708, USA

The NSF funded Assembling the Fungal Tree of Life (AFTOL) projectis based in four universities with worldwide participation (113 fungal systematists in 23 countries). The data organization and analyses of the 1500+ species of fungi for which 8 genetic loci
(i.e., 10 kb per species) will be sequenced will require new and efficient bioinformatic tools to access data and results shared by all participating researchers. A web-accessible SQL-database has been designed which facilitates data storage and sequence exchange between researchers, and with GenBank. The web-based database interface is built upon the Zope(c) application server and provides secure, real-time access to project data and results. Closely interacting with the database is a set of tools to automate the analysis of sequence data. Applications are written in the Python programming language and regularly use modules provided by the BioPython project. The system provides a through-flow of data from the sequencer to the database, and includes automated sequence quality checking, BLAST analysis, contiguous assembly, multiple sequence alignment, and phylogenetic analysis. The results of each step are evaluated and forwarded to their authors to allow further verification. A public website with limited access to the project data is available. The AFTOL database/ bioinformatics project uses only Open Source Software.


Allowing polytomies in Bayesian phylogenetic analyses, and an introduction to the next generation of collaborative phylogenetics software

Lewis, P. O. (1), Holder, M. (2) & Holsinger, K. E. (1)
(1) Department of Ecology and Evolutionary Biology, University of Connecticut, USA; (2) School of Computational Science and Information Technology (CSIT), Florida State University

This talk will be divided into two sections: the first covers recent theoretical work aimed at improving the robustness of Bayesian tree estimation, and the second presents the goals and initial efforts of a collaborative effort to expand the capabilities of phylogenetic inference methods. Within the past decade, Bayesian approaches to phylogenetic inference have become widely employed and accepted. Several simulation studies and surveys of empirical datasets have demonstrated that Bayesian and maximum likelihood approaches often agree with each other, but that the Bayesian analyses tend to indicate stronger support for many branches in the inferred tree. Current implementations of Bayesian methods only consider bifurcating trees; polytomies are only returned when multiple, incompatible bifurcating trees receive appreciable support. Explicitly allowing polytomies (by assigning them non-zero prior probability) can reduce the disagreement between Bayesian estimates of clade support and maximum likelihood bootstrapping. Reassuringly, considering polytomies does not dramatically reduce the power of Bayesian approaches to recover well-supported groups. The National Science Foundation ITR panel recently funded CIPRes, an effort to improve the performance and availability of phylogenetic methods. The software team includes the authors of Hy-Phy, MacClade, Mesquite, MrBayes, PAUP*, and POY as well as software engineers with expertise in integrating independent software. One of the projects goals is to produce freely available, state-of-the-art software for phylogenetic inference. The initial release demonstrates the flexibility of the system by implementing previously unavailable tree-searching methods. The algorithms were developed by computer scientists who are part of the CIPRes team, and have been implemented by writing new modules as well as tying together existing software. The CIPRes platform will transform how systematists analyze their data.


Where are we in assembling the fungal tree of life, classifying the fungi, and understanding the evolution of their subcellular traits?

Lutzoni, F. (1), Kauff, F. (1), Cox, C. J. (1), McLaughlin, D. (2), Celio, G. (2), Dentinger, B. (2), Padamsee, M. (2), Hibbett, D. (3), James, T. Y. (1), Baloch, E. (4), Grube, M. (4), Reeb, V. (1), Hofstetter, V. (1), Schoch, C. (5), Arnold, A. E. (1), Miadlikowska, J. (1, 6), Spatafora, J. (5), Johnson, D. (5), Hambleton, S.(7), Crockett, M. (5), Shoemaker, R. (7), Sung, G.-H. (5), Lücking, R. (8), Lumbsch, T. (8), O'Donnell, K. (9), Binder, M. (3), Diederich, P. (10), Ertz, D. (11), Gueidan, C. (1), Hall, B. (12), Hansen, K. (13), Harris, R. C. (14), Hosaka, K. (5), Lim, Y.-W. (3, 15), Liu, Y. (12), Matheny, B. (3), Nishida, H. (16), Pfister, D. (13), Rogers, J. (17), Rossman, A. (18), Schmitt, I. (8), Sipman, H. (19), Stone, J. (5), Sugiyama, J. (20), Yahr, R. (1) & Vilgalys, R. (1)
(1) Duke University, USA; (2) University of Minnesota, USA; (3) Clark University, USA; (4) Karl-Franzens-University, Austria; (5) Oregon State University, USA; (6) Gdansk University, Poland; (7) Agriculture and Agri-Food Canada; (8) The Field Museum, USA; (9) U.S. Dept. of Agriculture, IL; (10) National Natural History Museum, Luxembourg; (11) National Botanic Garden of Belgium; (12) University of Washington, USA; (13) Harvard University Herbaria, USA; (14) New York Botanical Garden, USA; (15) Curr. addr.: University of British Columbia, Canada; (16) The Institute of Physical and Chemical Research, Japan; (17) Washington State University, Pullman, USA; (18) U.S. Dept. of Agriculture, MD; (19) Botanischer Garten und Botanisches Museum Berlin-Dahlem, Germany; (20) The University of Tokyo, Japan

We present an overview of progress in molecular systematics of Fungi since 1990, and demonstrate that overlap among data matrices has been very low. As a result, many of the currently available data cannot be used in multi-locus analyses to infer fungal relationships on a large scale. We report here the results of four Bayesian analyses with complementary bootstrap assessment of phylogenetic confidence using neighbor joining, maximum parsimony, and Bayesian methods on: 1) combined two-locus data set (nucSSU and nucLSU rDNA) with 558 species representing all traditionally recognized fungal phyla (Ascomycota, Basidiomycota, Chytridiomycota, Zygomycota) and the Glomeromycota; 2) combined three-locus data set (nucSSU, nucLSU and mitSSU rDNA) with 236 species; 3) combined three-locus data set (nucSSU, nucLSU rDNA and RPB2) with 157 species; and 4) combined four-locus data set (nucSSU, nucLSU, mitSSU rDNA, and RPB2) with 103 species. The latter three analyses included only members of the Ascomycota and Basidiomycota. The four-locus analysis resolved multiple deep relationships within the Ascomycota and Basidiomycota that were not revealed previously, or that received only weak support values in prior studies. Based on these results and reanalysis of subcellular data, we also synthesize current knowledge regarding the evolution of septal features of fungal hyphae and present a preliminary reassessment of ascomal evolution. Topics to be discussed will include: How to improve the coordination of systematic studies on lichen-forming and allied fungi? Challenges and new developments in analyzing large-scale data sets. What can the project entitled "Assembling the Fungal Tree of Life" (AFTOL, funded by the National Science Foundation) can do for you?


Demonstration of MacClade and Mesquite

McMahon, M. M.
University of California, Davis, California, USA

Inference of character evolution is essential to reconstructing the tree of life. Character analysis allows, e.g., improvement in models of molecular sequence evolution that we apply to problems of tree inference, or testing hypotheses of correlations between molecular and/or morphological characters. The computer programs Mesquite (W.P. Maddison and D.R. Maddison, 2004) and MacClade (D.R. Maddison and W.P. Maddison, 2003) are powerful tools that allow evolutionary biologists to easily organize and study comparative data, graphically display the results, and creatively investigate alternate hypotheses of evolutionary pattern and process. In this presentation, the latest versions of these programs will be demonstrated. MacClade, as in earlier versions, performs parsimony reconstructions of characters and contains a graphical interface for manipulating trees. MacClade 4.0 also contains an extensive molecular sequence data editor, including tools for manual and automated alignment, amino acid translation, and calculation and display of consensus sequences. Mesquite, created as partner software to MacClade, is a platform-independent program whose capabilities grow as authors from throughout evolutionary biology add modules. For example, the PDAP module (Midford, Garland, and Maddison, 2002) calculates independent contrasts and the Rhetenor module (Dyreson and Maddison, 2004) performs phylogenetic morphometrics. Users see an entirely integrated environment, so there is no need to reformat data between analyses. The demonstration of Mesquite 1.0 will focus on a few of the many available features such as likelihood estimation of ancestral character states across sets of trees, detecting associations between trees involving evolutionarily linked lineages, and investigations of the coalescent process. For both programs, demonstrations will include tips on navigating context-dependent menus and information on import/export of data and graphics in various formats.

top