This directory contains all the working notes and predictions for
proteins in the CASP3 contest.

list-of-hits	A short list of all targets, one per line,
	indicating current status.

solved.ids	ids list (suitable for use with models.97 scripts)
		of all solved structures.

summaries	A collection of e-mail messages about the contest as a whole,
	or summarizing the status of predictions for many targets

make-all-joints	
	visit every target and make the alignments.
	This is useful to run after various library models have been created
	or rebuilt, to make sure that the predictions have all been updated.
	It probably should not be used now that some of the targets
	have been submitted!
	
possible.ids
	A list of PDB chains that have been considered as templates---useful
	for the models.97/scripts/apply-makefile to create fssp alignments
	or do other intensive work in the model library.  Taken from 
	the PRED= fields of make joint commands in the Makefiles.
	
many-possible.ids
	An early version of possible.ids, that contains more chains,
	but is missing some of the ones added later to Makefiles.

subdirectories:
    registration: contains the information about the group registrations.
    submit: contains the actual files submitted for each prediction,
	and the alignments on which they were based.
    networks-used: has copies of the neural nets used for secondary
	structure predictions, so they won't get lost in the mass of
	training and experimenting with the predictor.
    t*: There is a subdirectory for each target, which holds the
	working notes for that target.
	Each target should have t00xx.html, the html file from the casp3
	website; txx.seq, the sequence file in fasta format from the website;
	a Makefile that can be used to make all the computable files in
	the directory; and a README file with notes on which predictions
	look most promising and why a particular prediction was chosen.

	The Makefiles are generally made by copying from another
	target directory and updating as needed.  This allows the method
	to evolve over the summer, as we develop new techniques, but
	still records how the earlier results were created.
	
	The target directories will usually have a subdirectory for
	each chain that we are trying to find an alignment with, and
	there may be several alignments in such a subdirectory, usually
	created from the Makefile in the main t* directory.
    analysis:  contains a subdirectory for each target prediction 
	that is analysed in light of it's true structure