How to use GW to process ss.xpm file of secondary structures (ss)?

[lanselibai]

Sorry I am new to the GW. I have successfully installed it on PyCharm. What next should I do?

I tried import GromacsWrapper but got ModuleNotFoundError: No module named 'GromacsWrapper'

I also tried gromacs.g_hbond(s=TPR, f=XTC, g="hbond.log", hbm="hb.xpm", hbn="hb.ndx") but got NameError: name 'gromacs' is not defined

I want to import my ss.xpm file, then I can get an numpy array which contains all the ss records of all the residues for all the time frames. For example, ss[32][21] = 'E' means the 33th residue at 22th time frame has beta-sheet structure. Is this doable?

@orbeckst

[orbeckst]

You have to import it as

import gromacs

Reading a xpm file:

import gromacs.formats
xpm = gromacs.formats.XPM("./ss.xpm")

See the docs for XPM for more details how to work with the class.

[orbeckst]

To get the "33th residue at 22th time frame " do

resid = 33      # 1-based
frame = 22     # 0-based

resnum = xpm.yvalues[resid-1]
time = xpm.xvalues[frame]
state = xpm.array[frame, resid-1]

print("secondary structure for resid={0} at time {1:.3f} ps: {2}".format(resnum, time, state))

Note that the xpm.array contains the entries

{u'3-Helix',  u'5-Helix', u'A-Helix',  u'B-Bridge', u'B-Sheet', u'Bend', u'Chain_Separator', u'Coil', u'Turn'}

[lanselibai]

Thank you, I think GW works for python 2, not python 3. Can I ask if there is an alternative for python 3 as well? Or, can I use python 2 and 3 in the same time?

Or, what if I re-write your code into python3? How many files are involved in the xpm processing?

I am also interested in batch-processing rmsd/rmsf/gyration files, but I prefer python 3.

---

The following errors are for python 3 only:
Sorry, both import gromacs and import gromacs.formats could not work on both PyCharm and IDLE with SyntaxError: invalid syntax

  File "C:\Python36\lib\site-packages\gromacs\__init__.py", line 192, in <module>
    from . import fileformats
  File "C:\Python36\lib\site-packages\gromacs\fileformats\__init__.py", line 10, in <module>
    from .xvg import XVG
  File "C:\Python36\lib\site-packages\gromacs\fileformats\xvg.py", line 203, in <module>
    import gromacs.utilities as utilities
  File "C:\Python36\lib\site-packages\gromacs\utilities.py", line 275
    except OSError, err:
                  ^
SyntaxError: invalid syntax

[lanselibai]

Can I ask about the Chain_Separator? My protein has two chains, light chain (LC) from 1-214, heavy chain (HC) from 215-442. xpm.yvalues outputs 1 to 443, so there is one Chain_Separator. So is the 443th the Chain_Separator?

I just want to double confirm that. I also asked question on the Gromacs forum.
https://mailman-1.sys.kth.se/pipermail/gromacs.org_gmx-users/2018-February/thread.html
[gmx-users] Why "do_dssp" gives one more residue? ZHANG Cheng

[orbeckst]

We have not ported GW to Python 3 yet, see issue #44 . I'd like to but haven't had enough time. In principle it should not be hard because GW is pure Python. An initial attempt was PR #101, see comments there. If you work on making it work under Python 3 (while still supporting Python 2) I'd be happy to consider your contribution for inclusion – just submit a pull request (PR) on GitHub.

[lanselibai]

thank you for the effort, I will write my own python3.

[orbeckst]

Regarding Chain_separator: as far as I can tell you're right: it inserts an additional "residue". I have an example ss.xpm with two chains and each contains 374 residues:

>>> import numpy as np
>>> import gromacs.formats
>>> xpm = gromacs.formats.XPM("./ss.xpm")
>>> print(xpm.array.shape)
(10001, 749)
>>> separators = np.all(xpm.array == "Chain_Separator", axis=0)
>>> sep_resnum = xpm.yvalues[separators]
>>> print(sep_resnum)
[375]
>>> # I only have one separator so I am shortcutting and assigning it to 
>>> sep_indx = np.where(xpm.yvalues == sep_resnum[0])[0][0]

>>> # data for chain A
>>> ssA = xpm.array[:, :sep_indx]
>>> # data for chain B
>>> ssB = xpm.array[:, sep_indx+1:]
>>> print(ssA.shape, ssB.shape)
((10001, 374), (10001, 374))

[orbeckst]

If you fix the things that allow you to import on Python 3 and submit this as a PR then that could be a start.

[orbeckst]

I am also interested in batch-processing rmsd/rmsf/gyration files, but I prefer python 3.

You mean reading the output from the Gromacs commands in the form of XVG files? GW has the gromacs.formats.XVG class. But you can also use the np.loadtxt() function if you write XVG files without legends.

You can also run analysis of this kind in MDAnalysis, see

• https://www.mdanalysis.org/docs/documentation_pages/analysis/rms.html
• https://www.mdanalysis.org/docs/documentation_pages/overview.html?highlight=radius_of_gyration

[lanselibai]

for the MDAnalysis, I have some problems in installing it, and could not solve it at the moment. So writing my own one might be easier.

error: Microsoft Visual C++ 14.0 is required. Get it with "Microsoft Visual C++ Build Tools": http://landinghub.visualstudio.com/visual-cpp-build-tools

[orbeckst]

For MDAnalysis on Windows see https://github.com/MDAnalysis/mdanalysis/wiki/MDAnalysis-on-Windows

[lanselibai]

thank you. I do have ubuntu on VMwarePlayer on Win7. But I do not want to frequently use the ubuntu, as I need to copy and paste the files into it. I am writing my own python3 script to process the xvg/xpm files from gromacs.

[orbeckst]

Ok – I'll close the issue. Feel free to ask it to be re-opened (or re-open it yourself).