Contacts are a useful tool to analyse protein structures. They simplify the 3-Dimensional view of the structures into a 2-Dimensional set of contacts between its atoms or its residues. The representation of the contacts in a matrix is known as the contact map. Many protein structure analysis and prediction efforts are done by using contacts. For instance they can be useful for:

+ development of structural alignment algorithms [Holm 1993][] [Caprara 2004][]

+ automatic domain identification [Alexandrov 2003][] [Emmert-Streib 2007][]

+ structural modelling by extraction of contact-based empirical potentials [Benkert 2008][]

+ structure prediction via contact prediction from sequence information [Jones 2012][]

This code snippet will produce the set of contacts between all C alpha atoms for chain A of PDB entry [1SMT](http://www.rcsb.org/pdb/explore.do?structureId=1SMT):

AtomCache cache = new AtomCache();

StructureIO.setAtomCache(cache);

Structure structure = StructureIO.getStructure("1SMT");

Chain chain = structure.getChainByPDB("A");

// we want contacts between Calpha atoms only

String[] atoms = {" CA "};

// the distance cutoff we use is 8A

AtomContactSet contacts = StructureTools.getAtomsInContact(chain, atoms, 8.0);

System.out.println("Total number of CA-CA contacts: "+contacts.size());

The algorithm to find the contacts uses geometric hashing without need to calculate a full distance matrix, thus it scales nicely.

One can also find the contacting atoms between two protein chains. For instance the following code finds the contacts between the first 2 chains of PDB entry [1SMT](http://www.rcsb.org/pdb/explore.do?structureId=1SMT):

AtomCache cache = new AtomCache();

StructureIO.setAtomCache(cache);

Structure structure = StructureIO.getStructure("1SMT");

AtomContactSet contacts =

StructureTools.getAtomsInContact(structure.getChain(0), structure.getChain(1), 5, false);

System.out.println("Total number of atom contacts: "+contacts.size());

// the list of atom contacts can be reduced to a list of contacts between groups:

GroupContactSet groupContacts = new GroupContactSet(contacts);

See [DemoContacts](https://github.com/biojava/biojava/blob/master/biojava3-structure/src/main/java/demo/DemoContacts.java) for a fully working demo of the examples above.

[Holm 1993]: http://www.biomedcentral.com/pubmed/8377180

[Caprara 2004]: http://www.biomedcentral.com/pubmed/15072687

[Alexandrov 2003]: http://www.biomedcentral.com/pubmed/12584135

[Emmert-Streib 2007]: http://www.biomedcentral.com/pubmed/17608939

[Benkert 2008]: http://www.biomedcentral.com/pubmed/17932912

[Jones 2012]: http://www.ncbi.nlm.nih.gov/pubmed/22101153

A protein structure is determined by X-ray diffraction from a protein crystal, i.e. an infinite lattice of molecules. Thus the end result of the diffraction experiment is a crystal lattice and not just a single molecule. However the PDB file only contains the coordinates of the Asymmetric Unit (AU), defined as the minimum unit needed to reconstruct the full crystal using symmetry operators.

Looking at the AU alone is not enough to understand the crystal structure. For instance the biologically relevant assembly (known as the Biological Unit) can occur through a symmetry operator that can be found looking at the crystal contacts. See for instance [1M4N](http://www.rcsb.org/pdb/explore.do?structureId=1M4N): its biological unit is a dimer that happens through a 2-fold operator and is the largest interface found in the crystal.

Looking at crystal contacts can also be important in order to assess the quality and reliability of the deposited PDB model: an AU can look perfectly fine but then upon reconstruction of the lattice the molecules can be clashing, which indicates that something is wrong in the model.

This code snippet will produce a list of all non-redundant interfaces present in the crystal lattice of PDB entry [1SMT](http://www.rcsb.org/pdb/explore.do?structureId=1SMT):

AtomCache cache = new AtomCache();

StructureIO.setAtomCache(cache);

Structure structure = StructureIO.getStructure("1SMT");

CrystalBuilder cb = new CrystalBuilder(structure);

// 6 is the distance cutoff to consider 2 atoms in contact

StructureInterfaceList interfaces = cb.getUniqueInterfaces(6);

System.out.println("The crystal contains "+interfaces.size()+" unique interfaces");

// this calculates the buried surface areas of all interfaces and sorts them by areas

interfaces.calcAsas(3000, 1, -1);

// we can get the largest interface in the crystal and look at its area

interfaces.get(1).getTotalArea();

An interface is defined here as any 2 chains with at least a pair of atoms within the given distance cutoff (6 A in the example above).

The algorithm to find all unique interfaces in the crystal works roughly like this:

+ Reconstructs the full unit cell by applying the matrix operators of the corresponding space group to the Asymmetric Unit.

+ Searches all cells around the original one by applying crystal translations, if any 2 chains in that search is found to contact then the new contact is added to the final list.

+ The search is performend without repeating redundant symmetry operators, making sure that if a contact is found then it is a unique contact.

See [DemoCrystalInterfaces](https://github.com/biojava/biojava/blob/master/biojava3-structure/src/main/java/demo/DemoCrystalInterfaces.java) for a fully working demo of the example above.