You can now download from this page the programs and documentation relating to the Phylogenetic Regression (A. Grafen, 1989, The Phylogenetic Regression, Philosophical Transactions of the Royal Society Series B, 326, 119-157- this text is now available in PDF via my list of publications), and the full version of the appendix. If this page fails to work for you, please e-mail me to let me know.
The Philosophical Transactions paper included a shortened version of
the mathematical appendix, which contained definitions and theorems, but
no proofs. The full version was lodged in a manuscript archive, as detailed
in a footnote in the original paper. Now available here is the full
appendix, in pdf format. The typesetting has suffered in one or two
small details from its multiple transitions from Word 4 to Word 98 to Postscript
The Phylogenetic Regression was published on CRAN on 8th February 2014 as package "phyreg", version 0.7. It will take a couple of days to work its way through the system, though if you're in a hurry you can download the package source from here, and use
install.packages("pathname of the downloaded file", repos=NULL, type="source")
Soon, just install.packages("phyreg") will work.
As I no longer have access to SAS or GLIM, the R version is now the current version. It has all the previous versions had, and more.
(Version v0.5 was released on 15th March 2004, and version v0.6 on 23rd April 2005)
This new SAS implementation of the phylogenetic regression is very much easier to use, and has a number of additional features. Moving from data in an Excel spreadsheet, with a phylogeny expressed either as taxonomic variables or in newick-style, to getting F-ratios and p-values, is now a matter only of minutes. You will need the statistical package SAS (SAS Institute Inc, Cary, North Carolina).
Download three files and you're nearly there! You will need the manual (in PDF), and the program and examples (which are SAS transport files). The manual has details of the very straightforward installation procedure.
The version number is 0.7. It contains two bug fixes since the original 0.5. The first bug struck only when (i) a heights dataset was used AND (ii) not all species were used, in such a way that at least one higher node was omitted. The second bug struck when the taxonomic variables method was used, and the species were not in phylogenetic order. In both cases the bugs crashed the program, so that no analysis that did run was affected by them. Considerable testing showed that version 0.5 gives the same numerical results as the GLIM implementation in many circumstances. It is always possible there are more statistical bugs, but more likely there are presentational or informational infelicities. I will renumber after I know of more experiences, and probably some minor revisions. The manual has been very slightly updated to 0.7.
This SAS version will become the standard version. I no longer have access to the current version of GLIM, and future syntax changes in GLIM may leave PHYLO.GLM inoperative. The new version has a simpler interface, has better output, is more available to biologists, and deals with newick-style phylogenies.
The program itself is written in the statistical language GLIM,
distributed by NAG. This is a cross-platform
language available for personal computers and on UNIX machines, and mainframes
too. There are two versions of the program (phylo.glm and phylo8.glm) because
the new version of GLIM has a (poorly documented) failure of backward compatibility.
|Programs - these are text files, which you need to save to disk. NB On some systems it is necessary to increase the width of the input channel when reading these files e.g. $inp 11 120$ if phylo.glm is on channel 11.|
|Documentation - these are Adobe Acrobat pdf files, which can be viewed and printed more or less directly through modern browsers on nearly all platforms. If necessary, you can download the free Acrobat Reader yourself.|
|Data Files and Control Files allow you to do the example sessions yourself.They are text files, which you need to save to disk.|
As noted above, this GLIM version is unlikely
to be updated. GLIM is a slowly changing product, and so this is unlikely
to be of immediate importance to current users. For new users, however,
the SAS version should be the version of choice.
Alan Grafen 23rd April 2005
Back to Home Page