Propositions concerning NH treefiles

Rod Page dpage at udcf.gla.ac.uk
Fri May 10 06:07:12 EST 1996


Regarding Korbinian's first point, TreeView (http://taxonomy.zoology.gla.ac.uk/rod/treeview.html) 
supports both bootstrap values and edge lengths in a tree description using the approach taken by 
ClustalW. An internal node is labelled with it's bootstrap value, followed by an edge length.

The trees you gave can be combined to give

  (LungfishAu:0.1246,(LungfishSA:0.1338,LungfishAf:0.1324)100:0.0769,(
   (((Platypus:0.1441,Opossum:0.1302)100:0.0380,((Mouse:0.0566,Rat:0.0717)100
   :0.1017,(((Cow:0.0803,Whale:0.0935)100:0.0334,Seal:0.0848)100:0.0267,Human:0.1514)100
   :0.0355):0.0495):0.0865,((((Crocodile:0.1948,Bird:0.1546)100:0.0560,Sphenodon:0.2029)92
   :0.0239,Lizard:0.2302)72:0.0362,Turtle:0.1508)100:0.0581)100:0.0842,Frog:0.1664)99
   :0.0619);

where (Cow:0.0803,Whale:0.0935)100:0.0334 indicates that this group has a bootstrap value of 100 
and an edge length of 0.0334.

TreeView can display the bootstrap values on this tree, and at the same time draw the edges  
proportional to their length.

Exactly where bootstrap and other information (such as split numbers and decay indices) should go 
in relation to branch lengths is being discussed by a number of software authors,(e.g. 
Higgins at ebi.ac.uk,  joe at evolution.genetics.washington.edu, 
Gibson at embl-heidelberg.de,  r.page at bio.gla.ac.uk, mgouy at evomol.univ-lyon1.fr). The unpublished 
NEXUS standard suggests placing extra information in special comments after the edge lengths. It 
would be nice to come to some agreed standard that a range of programs would support.

Rod



More information about the Mol-evol mailing list