# What is the purpose of DSSP when using martinize2?

I am studying an IDP with a lot of helical secondary structure, and trying to find a trajectory that I can use as an ensemble that fits my SAXS data. My starting models are the output from the Robetta prediction server.

I am converting the IDP to a coarse-grained model so I can let the simulation run for ~$$10\mu s$$. In the process of converting it, I use martinize2. The python script takes the .pdb file, an output file name, and a pointer to the dssp binary (which I have looked everywhere for and haven't been able to find an 'official' source...).

martinize2 -f ../data/pdb/Robetta2.pdb -x cg.gro -o topol.top -dssp /usr/local/bin/dssp


If I run martinize2.py without specifying the -dssp parameter, the script runs fine and outputs my coarse grained model. If I specify the binary (I have probably tried 3 different versions of dssp), or specify -ss using the output from a dssp webserver, I will get different errors.

Am I okay to exclude the dssp parameter? Will my coarse grain model have the appropriate constraints for helical secondary structure from the PDB file? What is dssp even for in this context?

EXAMPLES OF ERRORS:

WARNING - unknown-input - Could not identify the modifications for residues ['ALA190'], involving atoms ['3059-3HB']
File "/home/nick/.local/lib/python3.8/site-packages/vermouth/dssp/dssp.py", line 214, in run_dssp
raise DSSPError(message.format(err=process.stderr, file=tmpfile_name))
vermouth.dssp.dssp.DSSPError: DSSP encountered an error. The message was DSSP could not be created due to an error:
empty protein, or no valid complete residues


...

File "/home/nick/.local/lib/python3.8/site-packages/vermouth/dssp/dssp.py", line 214, in run_dssp
raise DSSPError(message.format(err=process.stderr, file=tmpfile_name))
vermouth.dssp.dssp.DSSPError: DSSP encountered an error. The message was Error parsing PDB at line 1
Error trying to load file "/home/nick/GROMACS/projects/CG-4/dssp_in_i6lpkoe1.pdb"
Expected record CRYST1 but found ATOM


...

File "/home/nick/.local/lib/python3.8/site-packages/vermouth/dssp/dssp.py", line 557, in run_system
raise ValueError(
ValueError: The length of the sequence (13) does not match the number of residues in the selection (227).

• One thing I did not understand is why IDP has a lot of helical SS? Am I missing something. and the second thing is DSSP is an algorithm to predict the secondary structure. I guess you are just specifying your script that use this algorithm to predict the secondary structure. other algorithms such as STRIDE does a better job compared to DSSP. If you are satisfied with this, I can convert it into an answer. Jul 18, 2021 at 14:30
• It is a set of helices connected by disordered linkers. Imagine 8 helices with 10-15 AA random coil linkers connecting them. My question is more 'why' does martinize care about secondary structure? I have read that the martini force field cannot accomodate SS transitions, and I would like the helices in my all-atom model to remain as helices during the coarse-grain simulation. Jul 18, 2021 at 14:41
• Someone with more knowledge on martini force field should provide a better answer. But one requirement about martini FF is, the martini proteins which you simulate, the SS should be explicitly defined. This is done by DSSP. so why martinize care about secondary structure, because the residues which are identified as helical or beta sheet by DSSP, they have different parameters (like elastic force constant) compared to disordered linkers. Jul 18, 2021 at 14:52
• Related: cgmartini.nl/index.php/tutorials-general-introduction-gmx5/…. "The secondary structure of the protein influences both the selected bead types and bond/angle/dihedral parameters of each residue". I don't know the exact details of how they choose different beads, but they may just alter how coarse the mapping is depending on how much detail is needed to emulate the secondary structure.
– Tyberius
Jul 18, 2021 at 19:28

As Vasista notes, the parameters for beads can likely differ if its known to be part of a $$\beta$$-sheet or $$\alpha$$-helix.