This document is for OpenStructure version 1.1, the latest version is 2.7 !

Introduction to the mol Module

For the course of this tutorial, we assume that you have DNG up and running.

Loading and inspecting a protein structure

The code to load and save structures is not directly part of the mol module, but rather lives in a module dedicated to input and output of any kind of data: The io module. We will be using functions of this module to load structures.

One of the most commonly used file formats for macromolecular structures are PDB (Brookhaven Protein Databank) files. The official name for molecules stored in a PDB file is an entity and we decided to follow this convention in OpenStructure. You will hear this word all the time, but you can replace the word entity with molecule (or most of the time even protein) in your head.

To load a PDB file, simply type


This will load the fragment from the specified file ‘fragment.pdb’ and store the result in fragment. The LoadPDB() has many options, which, for simplicity will not be discussed here. If you want to know more about the function, type:


or read the online documentation.

The loaded structure is an instance of EntityHandle which offers a comprehensive interface to inspect an manipulate molecular structures. Now let’s inspect what we just loaded:

print len(fragment.chains), fragment.chains
print len(fragment.residues), fragment.residues
print len(fragment.atoms), fragment.atoms

As you can see, our fragment consists of one peptide chain of 12 amino acids and has 81 atoms in total. Now let’s examine our fragment in more detail. Enter the command

for residue in fragment.residues:
  print residue, 'has', len(residue.atoms), 'atom(s).'
  for atom in residue.atoms:
    print ' ',, atom.pos

This will group the atoms by residue. And, for completeness, we will first group them by chain, then by residues.

for chain in fragment.chains:
  print 'chain',, 'has', len(chain.residues), 'residue(s)'
  for residue in chain.residues:
    print ' ', residue, 'has', len(residue.atoms), 'atom(s).'
    for atom in residue.atoms:
      print '    ',, atom.pos

A protein fragment would not be complete without bonds: Let’s see what bonds we have in there:

for bond in fragment.bonds:
  print bond

From these short code examples we already see how the entity is structured: On one hand we have a hierarchy of chains, residues and atoms. On the other hand, we have bonds that form a network overlayed on the hierarchy. An important feature of entities is that we can always assume that the hierarchy is intact. You will never find an atom without residues, no residue can exist without a parent chain and chains belong always to an entity.

Let There Be Shiny Graphics

For visually inspecting the fragment, we now create a graphical representation of the entity. The graphical representation is completely separate from the EntityHandle class. This is on purpose. When writing processing scripts, usually no graphical representation is required and things would be slowed down without any reason. The following code will take our fragment and initialise a gfx.Entity, add it to the scene, and center the camera on it.

go=gfx.Entity("Fragment", fragment)

Now you will see the fragment in the 3D window.

Use the mouse to rotate, zoom in an shift the camera. Double clicking on an atom will center the camera on that atom. If you want to learn more about the gfx module, you are encouraged to read the gfx intro and the gfx documentation<ost.gfx.

Introduction to Views

Often during processing and visualisation of data, only parts of a protein structure are of interest. This realisation has had a major impact on the design of OpenStructure and is tied very deeply into the core of the framework. Subparts of structure are modeled as so-called EntityViews. You can think of them as a selection of chains, residues, atoms and bonds of an entity. A views has almost the same interface as the underlying entity, making it very easy to mix entity views with handles in Python due to the dynamic nature of the language. An algorithm that is written for entities will almost always (with some care) also work for EntityHandles. This is referred to as duck-typing (I don’ t care if it is a duck as long as it looks like a duck), a concept used all over the place in Python.  The view consists of one chain, one residue and two atoms. Again the same rule applies: No atom can be part of the view without it’s residue. In this example, no bonds are included, since there is at most one atom per bond in the original structure.

To familiarize yourself with the concept of views, we will use the fragment in the 3D window.

We will use several ways to select parts of our fragment:
  • By using a dedicated query language
  • By manually constructing a view

The Query Language

The first way to select parts of a structure is with a dedicated mini-language, called the query language. In the Python Shell, type


The code performs a selection on the fragment and assigns the resulting view to the selection of the graphical object. A green halo will be displayed around the selected parts (image in the middle).


As you can see the previous statement created a “full view”, containing all the chains, residues, atoms and bonds. To select lysine residues, type


As you can see (image on the right), the only lysine residue is now highlighted in the 3D window, because it was the only one matching the predicate “residue name must be equal to LYS”. Several such predicates can be combined with boolean operators such as and and or. To select residues with residue number 1 to 3, the following statement will do the job:

go.selection=fragment.Select('rnum>=1 and rnum<=3')

but this is very cumbersome. That’s why there is a shortcut to this statement. You can specify a range of values.


For a complete description of what you can do with the query language, have a look at the Queries.

Constructing Views Manually

Sometimes the query language Is Not Enough (TM). For these cases the construction of manual entities becomes neccessary. This is pretty straight forward:

ca=fragment.FindAtom('A', mol.ResNum(1), 'CA')
cb=fragment.FindAtom('A', mol.ResNum(1), 'CB')

The last step sets our constructed view as the current selection, displaying it in the 3D window. As you can see, C-alpha and C-beta of the first residue are not connected by bonds, even though both atoms are in the view. You have either to add the bond manually with


Or, as a very convenient shortcut view.AddAllInclusiveBonds() to add all bonds that have both bonding partners in the view.

Don’t forget to update the selection of the graphics object to see what view you have created.

Saving an Entity

Saving an entity (or a view) is a breeze:

io.SavePDB(fragment, 'full.pdb')

will save the full fragment. To save only the backbone atoms, we can first select the backbone atoms and then save it:

io.SavePDB(fragment.Select('aname=CA,C,N,O'), 'backbone.pdb')

That’s it for the mol module. Continue with part two of the tutorial.


Enter search terms or a module, class or function name.


Documentation is available for the following OpenStructure versions:

dev / 2.7 / 2.6 / 2.5 / 2.4 / 2.3.1 / 2.3 / 2.2 / 2.1 / 2.0 / 1.9 / 1.8 / 1.7.1 / 1.7 / 1.6 / 1.5 / 1.4 / 1.3 / 1.2 / 1.11 / 1.10 / (Currently viewing 1.1)

This documentation is still under heavy development!
If something is missing or if you need the C++ API description in doxygen style, check our old documentation for further information.