Statistical and structural analysis of 5913 protein kinase catalytic subunits:

Please direct all comments to kinase@t10.lanl.gov.  This web-site is 
riddled with errors, but has the benefit that it actually exists.  I
am working on things, but the problem is fairly difficult.

It is worthwhile to install something like 
chime  (http://www.mdl.com/downloads/downloadable/) 
or vmd (http://www.ks.uiuc.edu/Research/vmd/, although making vmd
work within netscape is a bit of a trick) to open pdb files
within your netscape browser.

This web-site derives from our group's efforts to make computationally
useful calculations of protein kinase function, based on currently available
sequence and structural information.  Much of it is correct, but the 
scale of the project makes it certain that large numbers of errors of 
all types are present.  Instead of fixing the errors before posting it on
the web, I tried to make it as easy as possible to check the validity of
any conclusions.  

The clickable-dendrogram is from "The protein kinase complement of the 
human genome" Manning, G; Whyte, DB; Martinez, R; Hunter, T; Sudarsanam, S
SCIENCE; DEC 6 2002; v.298, no.5600, p.1912, and is used with the permission
of the authors.

The labels are from either RefSeq (in which case they are clickable), or
from "Evolution of protein kinase signaling from yeast to man"
Manning, G; Plowman, GD; Hunter, T; Sudarsanam, S
TRENDS IN BIOCHEMICAL SCIENCES; OCT 2002; v.27, no.10, p.514-520.

The alignment of human kinase catalytic subunits was done entirely by
hand, using emacs and BioEdit (http://www.mbio.ncsu.edu/BioEdit/bioedit.html).
The sorting and alignment of kinase catalytic subunits from other species 
was made using hidden markov models derived from the human kinase alignment
and the HMMER suite of sequence tools (http://hmmer.wustl.edu/).  There are three
separate alignments.  The clikable structural motifs derive from a single,
hand-built alignment of the human kinase catalytic subunits.  The phylogenetic
trees were built from 8 separate alignments, chopped at 399 residues long.  The
sub-family alignments are un-touched output of clustalw, including one sequence
from the pdb.

Phylogenetic trees, entropies, and co-variation were computed from this
alignment with RIND and WEIGHBOR (http://www.t10.lanl.gov/rind/, 
http://www.t10.lanl.gov/weighbor/).

Homology models were built with MODELLER 
(http://www.salilab.org/modeller/modeller.html) from PKA, IRK, and PHK.
No particular effort has been made to make the homology model off of the
closest existing pdb file.

JPEG files of kinase homology models are made with VMD
(http://www.ks.uiuc.edu/Research/vmd/).  Backbone is colored in a 
smooth gradient from red to blue (N- teminus to C-terminus).  Sidechains
are shown as van der Waals spheres, color-coded according to residue
type (white- nonpolar; green - polar; red - acid; blue - base).

Alignments of complete protein sequences in each sub-family were 
constructed with CLUSTALW 
(http://www-igbmc.u-strasbg.fr/BioInfo/ClustalW/clustalw.html).

The kinase modeling project is funded by the United States Department
of Energy under contract # xxx through the LDRD program in a grant to
Ben McMahon.  Paul Fenimore, Bill Bruno, and Chang-Shung Tung have
contributed greatly to this effort.

Many things are color-coded and clikable.

I expect to update the website frequently in the next month.

Manuscript is in preparation.

Ben McMahon,
Los Alamos, NM 

1 Dec. 2003.