A) The biological assembly needs to be re-built and the atom coordinates of the asymmetric unit need to be rotated according to the instructions in the files. The information required to re-create the biological assemblies is available in both the PDB an mmCIF/PDBx files.

In PDB files the relevant transformations are stored in the *REMARK 350* records. For mmCIF/PDBx, the *_pdbx_struct_assembly* and *_pdbx_struct_oper_list* categories store the corresponding rules.

B) There is also a pre-computed file available that contains an assembled version of a structure. This file can be parsed directly, without having to perform rotation operations on coordinates.

BioJava contains utility classes to re-create biological assemblies for both PDB and mmCIF, as well as to parse the pre-computed file. The [BioUnitDataProvider](http://www.biojava.org/docs/api/org/biojava/nbio/structure/quaternary/io/BioUnitDataProvider.html) interface defines what is required to re-build an assembly. The [BioUnitDataProviderFactory](http://www.biojava.org/docs/api/org/biojava/nbio/structure/quaternary/io/BioUnitDataProviderFactory.html) allows to specify which of the BioUnitDataProviders is getting used.

Take a look at the method getBiologicalAssembly() in [StructureIO](http://www.biojava.org/docs/api/org/biojava/nbio/structure/StructureIO.html) to see how the BioUnitDataProviders are used by the *BiologicalAssemblyBuilder*.

This example in the next section loads the structure of the PBCV-1 virus capsid (PDB ID [1M4X](http://www.rcsb.org/pdb/explore.do?structureId=1m4x)). It consists of 16 million atoms and has one of the largest, if not the largest biological assembly that is currently available in the PDB. Needless to say it is important to change the maximum heap size parameter, otherwise there is no successfully load this. It requires a minimum of 9GB RAM to load (measured on Java 1.7 on OSX). You can change the heap size by providing the following startup parameter (and assuming you have 10G or more of RAM available on your system)

To load the pre-assembled biological assembly file directly, one can tweak the low-level PDB file parser like this

public static void main(String[] args){

public static void main(String[] args){

// This loads the PBCV-1 virus capsid, one of, if not the biggest biological assembly in terms on nr. of atoms.

// The 1m4x.pdb1.gz file has 313 MB (compressed)

// This Structure requires a minimum of 9 GB of memory to be loaded in memory.

String pdbId = "1M4X";

Structure bigStructure = readStructure(pdbId,1);

// let's take a look how much memory this consumes currently

System.out.println("Memory consumption after " + pdbId +

" structure has been loaded into memory:");

String mem = String.format("Total %dMB, Used %dMB, Free %dMB, Max %dMB",

r.totalMemory() / 1048576,

(r.totalMemory() - r.freeMemory()) / 1048576,

r.freeMemory() / 1048576,

r.maxMemory() / 1048576);

System.out.println(mem);

System.out.println("# atoms: " + StructureTools.getNrAtoms(bigStructure));

}

/** Load a specific biological assembly for a PDB entry

public static Structure readStructure(String pdbId, int bioAssemblyId) {

// pre-computed files use lower case PDB IDs

pdbId = pdbId.toLowerCase();

// we need to tweak the FileParsing parameters a bit

FileParsingParameters p = new FileParsingParameters();

// some bio assemblies are large, we want an all atom representation and avoid

// switching to a Calpha-only representation for large molecules

// note, this requires several GB of memory for some of the largest assemblies, such a 1MX4

p.setAtomCaThreshold(Integer.MAX_VALUE);

// parse remark 350

p.setParseBioAssembly(true);

// The low level PDB file parser

PDBFileReader pdbreader = new PDBFileReader();

// we just need this to track where to store PDB files

// this checks the PDB_DIR property (and uses a tmp location if not set)

AtomCache cache = new AtomCache();

pdbreader.setPath(cache.getPath());

pdbreader.setFileParsingParameters(p);

// download missing files

pdbreader.setAutoFetch(true);

pdbreader.setBioAssemblyId(bioAssemblyId);

pdbreader.setBioAssemblyFallback(false);

Structure structure = null;

try {

structure = pdbreader.getStructureById(pdbId);

if ( bioAssemblyId > 0 )

structure.setBiologicalAssembly(true);

structure.setPDBCode(pdbId);

} catch (Exception e){

e.printStackTrace();

return null;

}

return structure;

}

```