Re: [AMBER] Undocumented DSSP commands and torsional analysis

From: Daniel Roe <>
Date: Tue, 26 Aug 2014 11:33:01 -0600


On Tue, Aug 26, 2014 at 1:48 AM, Eugene Yedvabny <> wrote:
> 1) The newest manual lists 6 different ss types detected by DSSP algorithm,
> but looking at the source code, there is support for the 7th 'Bend' type.
> In my secstruct outputs I don't see a '7' type in either the raw or the
> averaged outputs. Can I then assume I have no bends in the peptide, or is
> that code still experimental and perhaps not reliable?

Support for 'bend' was added along with other fixes to the DSSP code
as part of AmberTools 14 update 6. As long as your version of cpptraj
is >= 14.05 (i.e. updates were applied and the code was recompiled
with no error) the bend calculation should be present and working (I
ran *many* tests prior to releasing the update). If you have reason to
believe the bend calculation is not working properly let me know. You
can test that the DSSP algorithm in your cpptraj binary is correctly
assigning bend using the attached PDB and output files, and using the
following input:

dssp assignout assign.4.dat

> 2) Is there any way to parse the *type* of a Turn across a group of
> residues? For example, secstruct stats show a Turn on residues 6, 7 and 8,
> which suggests an alpha-turn 5-9. But the h-bond analysis reveals that
> there's a mixture of 5-8 and 6-9 turns in addition to 5-9 in the ensemble.
> Is there a way to refine reported 'T' into 'T4', 'T5', or 'T6' for the
> different lengths of turns short of modifying the source?

No, but it should be possible to script something up using the SS v
frame output from 'out'. Also, be careful when you are comparing DSSP
results directly to what you get from 'hbond', since the DSSP hydrogen
bond criterion is different than what is used in the 'hbond' command.
The former uses an energy-based cutoff that takes distance and angle
into consideration at the same time, while the latter uses simple
geometric cutoffs that must *both* be satisfied. So something that you
may consider hydrogen bonded based on simple geometric considerations
may not appear to be so in DSSP or vice versa.

> 3) DSSP is intrinsically based off h-bonds. Is there any algorithm in
> cpptraj that would perform a comparable analysis on the basis of dihedral
> angles? I am trying to detect if there are any non-h-bond-stabilized
> structural motifs in the peptide.

Not yet. If you script something up however you may find the
'multidihedral' command useful for calculating a large number of
common dihedral types.


> Thank you,
> Eugene Yedvabny
> _______________________________________________
> AMBER mailing list

Daniel R. Roe, PhD
Department of Medicinal Chemistry
University of Utah
30 South 2000 East, Room 307
Salt Lake City, UT 84112-5820
(801) 587-9652
(801) 585-6208 (Fax)

AMBER mailing list

Received on Tue Aug 26 2014 - 11:00:03 PDT
Custom Search