PAL Release History

PAL was first publicly distributed in September 1999. This and all subsequent releases can be downloaded by clicking on the corresponding version number.
1.2 Codon models implemented (rate matrix, alignment, codon translation methods). Utility functions for alignments (score function etc.). Splits, split systems and corresponding tree distances (Robinson-Foulds). Extended utility functions for trees (incl. Node factory). Pareto distribution. Classes for "measurably evolving populations" (mep package). (April 2001).
1.1 Improvement of coalescent package: generalized skyline plot, new const.-to-exp. growth demographic model, new features in IntervalsExtractor, new classes DemographicTree/ClockTree. Reorganization of math and statistics packages: new binomial, error and gamma functions in math package, and bootstrap statistics (BIAS, VAR, CI etc), gamma, exponential, chi-square, and normal distributions with pdf, cdf, and quantiles in statistics package. New features in TreeUtils and NodeUtils, allowing for a smaller SimpleTree class. SUPGMA tree reconstruction and distances. Nucleotide-to-amino acid translation class. More GUI classes, e.g., for circular trees. (December 2000).
1.0 Tree and Node classes turned into Interfaces implemented by SimpleNode and SimpleTree and controlled by TreeUtils and NodeUtils classes. New ParameterizedTree: Clock-trees with dated tips. Much better accuracy and vast speed-ups in algorithm for ML optimization of branch lengths for UnconstrainedTree, ClockTree, DatedTipsClockTree (optimization delivers now "real-world" performance). New tree comparison tests (e.g., Shimodaira-Hasegawa). New protein models (VT, WAG, CPREV). New classes for demographic models, coalescent simulation and coalescent priors. Package names changed (clearer longer names), some classes reorganized, and two new packages introduced (coalescent and util). Sorting class. Macintosh compatibility (some file names changed). (July 2000).
0.9 Formation of PAL Development Core Team. PAL converted to pure library (command applications moved into the Vanilla package). More efficient multivariate minimization classes (non-linear conjugate gradient algorithms, line function with boundaries, conjugate direction method with boundaries etc.) and improved stop criteria. Automatic model selection among amino acids matrices. Inclusion of the VT rate matrix. Estimation of model parameters without a tree. Several smaller bug fixes and changes (alias mapping in Tree reversed plus subsequent changes in TreeDistance and LikelihoodValue). (April 2000).
0.8 All classes are now fully documented using JavaDoc comments. Source code complies to Java 1.0 language standard to allow compilation into native code. Classes for evaluation of evolutionary hypotheses (chi-square and likelihood computation) and interfaces for functions and of Parameterized are reorganized. Detail improvements in a number of other classes. Introduction of clock-constraint on trees. Some minor bug fixes. New platforms: direct support for Windows and Unix-native (includes instructions for compilation into native code using the GNU Compiler for Java). New applications: UPGMA (constructs UPGMA tree), STRIPSITES (strips sites from an alignment according to a given order). LSTREE and MLTREE now enable the estimation of clock-like branch lengths. (February 2000).
0.7 Dog icon. Much improved classes for formatted input and output. PHYLIP sequential format added. Restructuring of alignment package (e.g., removal of TypedAlignment, changes in SitePattern, automatic data type detection). Introduction of getInstance() methods in DataType and RateMatrix. Objects for computing tree-induced distances and (weighted) squared distances between distance matrices. Class for computing least-square branch length estimates. CUI frontend now in a single common class for all CUI programs. New applications: EVOLVE (generation of artificial data sets), LSTREE (least-squares estimates of branch lengths), JUMBLE (randomizes sequence input order). (December 1999).
0.6 Complete reorganization of package layout and all classes. Main new classes: ML evaluation of trees, estimation of model and branch parameters, global optimization (genetic algorithm), likelihood ratio and Kishino-Hasegawa-Templeton tests, random number generator, numerical derivatives of vectors. New application: MLTREE (computation of likelihoods given a tree and estimation of model parameters and branch lengths). Improvement of MLDIST program. (November 1999).
0.5 First public release. (September 1999).
- Project start. (January 1999).

Back to main page

Last modified: April 8, 2000