Frames and Atom Groups

This part shows how to use AtomGroup in place of Frame.

Trajectory Frames

Frame instances store only coordinate and some frame related data. For each frame data, you will a different frame instance:

In [1]: from prody import *

In [2]: dcd = Trajectory('mdm2.dcd')

In [3]: dcd
Out[3]: <Trajectory: mdm2 (next 0 of 500 frames; 1449 atoms)>

In [4]: frame0 = dcd.next()

In [5]: frame0
Out[5]: <Frame: 0 from mdm2 (1449 atoms)>

In [6]: frame1 = dcd.next()

In [7]: frame1
Out[7]: <Frame: 1 from mdm2 (1449 atoms)>

These Frame instances are different objects:

In [8]: frame0 is frame1
Out[8]: False

When you are not referring to any of these frames anymore in your code, Python garbage collector will free or reuse the memory space that was used by those frames.

Linking Atom Groups

When trajectory is not linked to an AtomGroup (using link()), Frame and AtomGroup objects share the same coordinate data. Let’s see how this works:

When an AtomGroup is linked to the trajectory as follows, things work differently:

In [9]: pdb = parsePDB('mdm2.pdb')

In [10]: pdb
Out[10]: <AtomGroup: mdm2 (1449 atoms)>

In [11]: dcd.link(pdb)

In [12]: dcd.reset()

We get Frame instances in the same way:

In [13]: frame0 = dcd.next()

In [14]: frame0
Out[14]: <Frame: 0 from mdm2 (1449 atoms)>

In [15]: pdb.getACSLabel()
Out[15]: 'mdm2 frame 0'

Note that the active coordinate set of the AtomGroup and its label will change when we get the next frame:

In [16]: frame1 = dcd.next()

In [17]: frame1
Out[17]: <Frame: 1 from mdm2 (1449 atoms)>

In [18]: pdb.getACSLabel()
Out[18]: 'mdm2 frame 1'

Now the key difference is that the Frame instances are the same objects in this case:

In [19]: frame0 is frame1
Out[19]: False

As you see, a new frame was not instantiated. The same frame is reused and it always points to the coordinates stored in the AtomGroup. You can also make Selection instances that will point to the same coordinate set. This will allow making a more elaborate analysis of frames. For an example see Trajectory Analysis II.