Wednesday - wrightscape runs and various other things

pdg-control

  • plots for changing costs
  • Report out to Training Problem II group.

ggplot note

defining functions that return a ggplot object can be tricky. If you compute some objects in the plot function (stats etc), those stats are not stored by the object (due to lazy evaluation, usually a very nice time-saving feature), so the plot cannot be produced unless the function returns those things to the global environment (or from wherever the plot will be printed/evaluated). Can return them to the global env with “<<-” assignment, but seems poor form.

Evolution

power in phylogenies

  • Hessian reply.  manuscript reply.
  • Select wrightscape examples.

wrightscape

  • 1530089 series: bm vs a2 on parrotfish tree. looks like close & SH.y reject brownie, open almost does. kt almost does.
  • 7a347b0 series: bm vs a1 on parrotfish tree: (missing lr line).  More power, smaller diffs.  Kt maybe? must rerun these!
  • 4335204  bm vs a1 with the lines.

prot.y sig. open sig in reverse

  • cce5a5a/93a8630 series: labrid tree, intramandibular division, bm vs a1   (more power on larger tree, go for finer division between models?)   only bodymass rejects.  (cce5a5a missing lr line).

  • 8382283 now running labrid bm vs a2.

KT significant 

closing: boarderline
closing: boarderline
bodymass sig
bodymass sig
open: kinda not really
open: kinda not really
  • open & prot.y significant in reverse
  • 6befae4 Check out open and kt on a2 vs bm when regime=two_shifts.
Prot
Prot
open
open
kt
kt

bodymass are significant.  Close is significant reversed (funny pattern at a2 anyhow).

Make sure to run against standard ouch too!

Warning Signals

  • Alan Manuscript
  • Reading: re data sources (Dulvy et. al. 2003)
  • (Le Quesne & Jennings, 2012)
  • Example decisions under uncertainty of crash

Reading

  • (Molloy, 2011)
  • And on NC clause (Carroll, 2011), (Hagedorn et. al. 2011)
  • Fiction: Non-commerical license means no one can make money from it. “I don’t want to give some entrepreneur a free lunch”
  • Fact: NC means the journal (by holding the copyright) retains the right to sell your work to pharmaceuticals, teachers, researchers interested in text-mining,  (rather than those groups having free use).
  • Data reviews – F1000?

[flickr-gallery mode=“search” tags=“phylogenetics, PDG_Control” min_upload_date=“2011-12-07 7:00:37” max_upload_date=“2011-12-08 12:23:37”]

References