DSSP Implementation

[William Glass]

Hi everyone,

I'm looking to characterise loss or gain of secondary structure over a simulation.

I was wondering if MDAnalysis has dssp implemented in it or if this is something that needs to be worked on?

I guess the general idea would be to produce a .pdb file at every N frames and pass this to dssp.

Any help would be great!

Will

[Oliver Beckstein]

Hi Will,

Welcome the MDAnalysis list.

On Sep 25, 2018, at 7:19 AM, 'William Glass' via MDnalysis discussion <mdnalysis-...@googlegroups.com> wrote:

Hi everyone,

I'm looking to characterise loss or gain of secondary structure over a simulation.

I was wondering if MDAnalysis has dssp implemented in it or if this is something that needs to be worked on?

We don’t have this implemented. There’s an open issue https://github.com/MDAnalysis/mdanalysis/issues/1612 

I guess the general idea would be to produce a .pdb file at every N frames and pass this to dssp.

That’s a common approach (for instance, the MDAnalysis.analysis.hole module uses it). 

Even better would be a “pydssp” package that provides a Python interface to the DSSP algorithm. If I had the time, I would use the open source sources for DSSP and try to write  a simple DSSP library (either C++ or Cython) with a generic Python interface that would be easy to use from MDAnalysis (and any other code that wants to use it).

But in the meantime, something like

protein = u.select_atoms(“protein”)

pdbfile = “temporary.pdb”

data = []

with mda.Writer(pdbfile, “w”) as W:

   for ts in u.trajectory[::step]:

       W.write(protein)

       secstructure = run_dssp(pdbfile)

       data.append( (ts.frame, ts.time, secstructure) )

where run_dssp() should be a function that calls dssp (e.g. with subprocess.run()) and returns the secondary structur information as a suitable data structure. We then store it in `data` as a time series and then you have to process `data`. (You can also create proper temporary files that get cleaned up automatically but I just put down the bare bones.)

Oliver

Any help would be great!

Will

--

Oliver Beckstein

orbe...@gmail.com * orbe...@gmx.net