Moved Interval* and IntPair* classes to tools package

Now removed variable and put back code that shouldn't have been commented out

Commented out some unused variables to get rid of warnings

Refactoring of inconsistently named methods in RIGraph

Reorganised the project with a src folder for java source files.
Added a jars dir with all jars needed for the project.
Added .project and .classpath pointing to relative path of jars. 
The project should now work out of the box after a check-out with eclipse. No need to setup external jars or anything.

fields changed to private

fields changed to private

all fields to private

some methods modifiers were changed from protected to public, since classes of the sub package rely on those

all classes for evolution, correlation function, neighborhood string converter etc.

full package, with all classes

minor changes

final version, might still contain bugs, but seems consistent so far

the older version still had a bug in the setNbhStringMaps(String,String) method

a class that gets neighborhood Strings and queries them

helper class

Major commit: 
- change of the implementation of Pdb (interface remains backwards compatible). Now data stored in a proper object-oriented way. Pdb not abstract anymore, a protein model can be created by using one of the constructors
- added many tests and test files for Pdb (and merged some already existing tests from the default package)
- removed obsolete classes related to msd
- bug fix: PdbfilePdb now properly reads files with alt codes
- all tests passed

Fixed +1 issues

class to estimate the correlation function to a given error function, deals onlz with Population arrays

new class to deal with arrays of Individuals class, provides methods to generate new arrays of Individuals

major bugs removed and documentation added, couple of method removed to new classes, i.g. Population class and CorrelationFunction class

fixed randomSets method

new class for some evolution

distance map and contact map error function

Move the scoring methods to their own class as statics. Did a bit of refactoring.

One more command line argument: contact percent sampling

Changed scoring: from sum(upi-ui)2 to sum of max(0,upi-ui).
New methods to write subsets to files in table (tab delimited) format.
Now main() writes out all scores/subsets to two files instead of only outputting best, worse and ensemble of bests.
Moved ScoreSet to its own file, new static methods to read subsets from db.

New method to get an ensemble RIG from a group of best scoring sets.

Changed the scoring so that is an error function: now it is the average sum square upper bounds deviation from the contact map. A perfect fit to the contact maps is err=0.
Cleaned up, refactored and documented.

New experimental class Distiller

Now printing statistics for both lower and upper violations.

Fixed bug: this time really fixed bug from last revision. Now references of the bounds arrays are properly handled (hopefully!). Tests seem correct: now the output models when metrizing seem not to be correlated anymore

Fixed bug: was reusing the same bounds after every metrization in reconstruct(), now restarting from the initial bounds for each metrization.
In metrize() final sampled matrix is now always reinitialising seed.

Now printing restrains viols and bounds viols.

Now Reconstructer can also take RIGraphs with "unobserved" residues (i.e. with missing nodes).
Model pdb files that Reconstructer creates now contain strictly only CA atoms.

New classes Reconstructer and ModelPdb. Reconstructer is a one-stop solution for reconstructing from contact maps (RIGraph) using BoundsSmoother and Embedder. At the moment doesn't support reconstruction of contact maps with non-observed residues.

Made Bound[][] bounds a member, thus changing the signature of most methods. Refactored getBoundsAllPairs(bounds) to better name computeTriangleInequality: now it uses as input the bounds member variable and updates it as output.

Fisrt full implementation of metrization 

Some big changes: now internally we use the data structure Bound[][] for all steps. We only use graphs for when we need to calculate shortest paths.
Define some static functions for printing info.
Made the scaling method an option passed as an argument to embed()
Extracted to separate file the Bound class.
Drafted an implementation of metrize

Introduced scaling of the embedded coordinates (first good embedding ever!!)
Make the random generation more consistent. Also introduced for debugging a fixed seed.

Fixed in BoundsSmoother: now the lower bound will be set to the triangle inequality value only if the newly found lower bound is bigger than the hard spheres bound.
Now we also introduce contiguous CA distance constraints (3.8)
In main now checking for violations to the full bounds matrix.
New method in Pdb for getting distance matrix as Jama object

New method in Pdb to set atoms for all coordinates of a given atom. Used in BoundsSmoother and Embedder mains.
Changed slightly how parameters are passed in BoundsSmoother.

Lower bounds bounds-smoothing now also implemented (using Dijkstra's shortest path algorithm and offsetting to avoid negative weights).

First implementation of BoundsSmoother. Does only bounds smoothing through triangle inequality for the upper bounds (for lower bounds just takes a hard-spheres distance value)

First implementation of the embedding part of the EMBED algorithm by Havel: from an exact complete distance matrix gives 3d coordinates for the embedding.