Changes for PDAP:PDTREE for version 1.08
Note that starting with this version, the Mac OSX release will use the DMG archive
format, rather than the SIT format.
- Added support for correcting the degrees of freedom on screen 9. This change
reflects that this screen reports p-values, which are calculated from the t-statistic
reported for the regression. If you have set verbose mode, you will be asked about
this each time to enter screen 9. Otherwise, the program will now ask exactly once for
each tree during a session. The command to manually change the degrees of freedom
has also been added to the menu for this screen. Note that:
- the raw t and F statistics are correct and unchanged in this version.
- the DOS version of PDAP never reported a p-value for the t statistic and therefore
needs no updating.
- if you use the "Felsenstein correlation" calculator for a tree and two characters to generate
a p-value, it will always treat polytomies in the most conservative way: subtract one degree of
freedom for each branch collapsed into a polytomy. Because of the contexts that this calculator
is likely to be used, it is not practical to ask the user for each tree that contains polytomies.
Last update of this page 15 December 2006
- Pagel's branch length transform 'broke' polytomies in trees loaded from PDI
files. The problem was caused by the PDI file reader retaining zero length branches
that PDAP uses to represent polytomies. The reader now collapses zero length
branches into Mesquite-style polytomies.
- Column 9 of the FIC file incorrectly duplicated column 8 (raw node heights)
instead of reporting the corrected node heights.
- An problem with the variance of contrasts returned by as a "number for tree",
for example in a tree legend has been fixed. The number was incorrectly calculated
and was different from the value displayed on screen 9 of the PDAP:PDTREE diagnostic
charts, which is correct. The discrepancy has been fixed.
- The log10 character transform was incorrectly reporting some data as nonpositive.
- The unassigned branch length checker was reporting problems with generated trees,
such as the Yule trees in Part 6 of the guided tour. The only unassigned branch length
was in the branch below the root. Since PDAP never uses this branch, the checker is
now set to ignore it.
- Problems in saving large (493 tip) trees to a PDI file have been fixed.
- The guided tour files have been updated and no longer report being of an older version.