NMR Subchain Search, circa 1/98

You should be able to compile this by doing
	make nmrc
also do
	cc -o labels-to-pdb labels-to-pdb.c

I've compiled nmrc on Sun, SGI, PC-Linux and NT with no problems.
However, it seems that compiling with optimization on some machines
leads to substantially different results.  I've yet to track down the
reason. Any help on this will be appreciated.

You can try it with the command:
	nmrc -in znf-chains-14-28 -e 0.025 -b 1 -one

This should produce (after some long time e.g. >20min) a number of files
of the form
	nmr-chain-x-y 
which have torsion angles (and error) for solutions, as well as a
final nmr-chain-14-28.labels which has positions for all the solutions
of the "whole" chain

You can turn the labels file into a PDB file by running
	labels-to-pdb znf-14-28-pdb nmr-chain-14-28.pdb
This just substitutes the positions into the original znf-14-28.pdb
file.  You can then compare the original one to that produced by the
search.  You should have a way of displaying PDB files (if you don't,
get Rasmol).  BTW, the last ten or so atoms in the PDB file are
pseudo-atoms (marked by # in the name), if they give you trouble,
flush them (but they are needed when labels-to-pdb runs).

	The Code:

The command interface is in nmr-subchains-main.c

Basic support functions, file reading as well as basic chain
kinematics is in nmr-subchains-basic.c

The functions that manage the torsional angle search are in
nmr-subchains-search.c

The error and derivative computations as well as the interface to the
minimizer is in nmr-subchains-error.c

The "condense" process that tries to merge solutions is in
nmr-subchains-merge.c.

The minimizer is in cfsqp.c and qld.c.  See the cfsqp_* files for
information on it.  You should (sometime) contact the authors via
E-mail to get their official permission to use it.  See their Web
page.

Tomas
TLP@AI.MIT.EDU
