2
Ziheng Yang, 27 January 2003
5
This folder contains files for the local-clock likelihood analysis of Yoder
9
The clock models are re-written. Note that the fossil calibration
10
points are specified using the symbol '@' in the tree file; the symbol
11
is used in the same way as the more familiar symbol ':' for branch
12
lengths. TipDate models are implemented using @ in the sequence name
13
to specify the date of determination of the sequence. Look at the
14
examples/TipDate folder for an example. Local clock models are
15
specified by assigning rates to branches, using the symbol # to label
16
the branch/node. The symbol $ is used as a clade label; $ stands
17
for the capital greek symbol \Delta, which looks somewhat like a clade.
19
Use clock = 2 for local clock models. For models of multiple genes,
20
clock = 3 represents the "combined analysis" in Yang & Yoder (2003).
21
The program assumes the same set of species and the same phylogeny at
24
clock = 1: global clock, deals with TipDate with no or many fossils,
25
ignores branch rates (#) in tree if any.
26
= 2: local clock models, as (1) above, but requires branch
28
= 3: as (2) above, but requires Mgene and option G in sequence
34
The following describes specifications in baseml.ctl to duplicate the
35
ML results in tables 4 and 5 of Yang and Yoder (2003). Note that the
36
ML analysis uses the ingroup sequences only, and the assumed ingroup
37
tree is rooted. In the Bayes analysis of that paper, Thorne's program
38
(estbranches) requires an outgroup to root the ingroup tree. The
39
"JeffNode" node numbering corresponds to the the node numbering by
40
Jeff Thorne's Bayes MCMC programs divtime5b and multidivtime, used in
41
our paper. My output is correct only if you do what we did, that is,
42
only if you use the outgroup in Jeff's estbranches program and do not
43
use the outgroup sequences in the baseml ML analysis. Note that the
44
node numbering in Tables 4 and 5 of that paper is according to Jeff's
45
programs divtime5b and multidivtime.
48
JC69 analysis in Tables 4 and 5
49
===============================
51
Columns b, c, d. Use optoin GC on the first line of the sequence data file
56
The tree in the tree file has seven @ specifying seven calibration
57
dates. Use Mgene = 1 for separate analysis. Other options are
65
These options should generate the three columns in one analysis. The
66
option clock = 1 has precedence over the branch rate labels # in the
67
tree file. So clock = 1 above means that the model is a global clock
68
and the rate labels # are ignored.
70
Columns b, c, d in table 5 (local clock analysis). If you choose
71
clock = 2 in the above, you should recover results for the local clock
74
Column i in table 4 (combined global clock analysis under JC69) is
75
produced by the following options:
83
Column i in table 5 (combined local clock analysis under JC69) is
84
produced by changing clock = 3 in the above.
89
F84G analysis in Tables 4 and 5
90
===============================
92
Use the following to specify the gamma model.
95
alpha = 0.5 (or any other initial value)
98
Columns f, g, h in table 4 (global clock analysis). Use optoin GC on
99
the first line of the sequence data file. The tree in the tree file
100
has seven @ specifying seven calibration dates. Use Mgene = 1 for
101
separate analysis. Other options are
106
alpha = 0.5 (or any other initial value)
109
Columns f, g, h in table 5 (F84G local clock analysis). Change clock
112
Column j in table 4 (combined global clock analysis under F84G) is
113
produced by the following options:
123
Column j in table 5 (combined local clock analysis under F84G) is
124
produced by setting clock = 3 in the above.