becksteinlab / mdpow Goto Github PK

View Code? Open in Web Editor NEW

25.0 25.0 10.0 756.99 MB

Calculation of water/solvent partition coefficients with Gromacs.

Home Page: https://mdpow.readthedocs.io

License: GNU General Public License v3.0

Python 99.80% Shell 0.20%

free-energy gromacs gromacswrapper molecular-dynamics python science

mdpow's People

Contributors

Stargazers

Watchers

Forkers

leelasd chemlove vod555 tsenapathi zhu-liu alescoulie asclepiusinformatica bbyun28 rnaimehaom a-ws-m

mdpow's Issues

customize initial box dimensions

Add a distance parameter to the config file to set the initial minimum distance between solute and rhombic dodecahedral box. (Override the defaults.)

move cfg 2 yaml conversion script

Put the conversion script to the script directory and install it as mdpow-cfg2yaml.py.

(Note: possibly need to add some code to setup.py to make it install and package.)

update links in docs

Change old SBCB links in setup.py and docs to GitHub.

setting mdrun.maxthreads to anything but None raises AttributeError

In the new input file (yaml parser), the following raises AttributeError:

mdrun:
  maxthreads: 8

Traceback:

Traceback (most recent call last):
  File "/nfs/homes2/oliver/.local/bin/mdpow-equilibrium", line 6, in <module>
    exec(compile(open(__file__).read(), __file__, 'exec'))
  File "/nfs/homes2/oliver/Projects/Methods/MDPOW/scripts/mdpow-equilibrium", line 71, in <module>
    S = equilibrium_simulation(cfg, opts.solvent, dirname=opts.dirname)
  File "/nfs/homes2/oliver/Projects/Methods/MDPOW/mdpow/run.py", line 216, in equilibrium_simulation
    success = wrapper(runMD_or_exit, S, "MD_relaxed", params, cfg)  # note: MD_relaxed!
  File "/nfs/homes2/oliver/Projects/Methods/MDPOW/mdpow/restart.py", line 211, in dummy_protocol
    success = func(*args[1:], **kwargs)
  File "/nfs/homes2/oliver/Projects/Methods/MDPOW/mdpow/run.py", line 137, in runMD_or_exit
    nt=cfg.get('mdrun','maxthreads'),
  File "/nfs/homes2/oliver/Projects/Methods/MDPOW/mdpow/config.py", line 121, in get
    if value.lower() == "none":
AttributeError: 'int' object has no attribute 'lower'

(This was also reported by @iorga .)

scripts for analyzing cyclohexane simulations

The current develop branch does not install a script similar to mdpow-ghyd that computes the solvation energy in cyclohexane.

(There's also no such script of octanol so we really should have a general mdpow-solvationenergy --solvent=<SOLVENT> script.)

install gromacs on travis

CI on travis is now running (#24) but for integration tests (part of #23) we need a working Gromacs installation.

We would like to test at least:

Gromacs 4.6.6
Gromacs 5.1.2

See also Becksteinlab/GromacsWrapper#39 – any solution from GromacsWrapper can be applied here.

script mdpow-pcw needed for calculating cyclohexane/water partition coefficient

The behavior should be similar with those of mdpow-pow. I have created an initial version by copying mdpow-pow and replacing octanol by cyclohexane and oct by cyc (Attention, we have Gcyclo.fep). Committed with SHA: 30ab593.

Currently it fails with the following error:

mdpow       : INFO     Analyzing directory '070'... (can take a while)
mdpow       : INFO     Reading water data '070/FEP/water/Ghyd.fep'
mdpow.fep   : INFO     Solvation free energy calculation for molecule UNK in solvent water.
mdpow.fep   : INFO     Base directory is '/gem/iorga/projects/sampl5_sims/070/070'
mdpow.fep   : INFO     Using setup directories under 'FEP/water': {'vdw': 'FEP/water/VDW', 'coulomb': 'FEP/water/Coulomb'}
mdpow.fep   : INFO     Default checkpoint file is '/gem/iorga/projects/sampl5_sims/070/070/FEP/water/Ghyd.fep'
mdpow       : INFO     Reading cyclohexane data '070/FEP/cyclohexane/Gcyc.fep'
mdpow       : CRITICAL Running analysis in directory '070' failed
mdpow       : ERROR    Catastrophic problem occurred, see the stack trace for hints.
Traceback (most recent call last):
  File "/home/chem/iorga/temp_ftp/mdpow-pcw", line 232, in <module>
    force=opts.force, stride=opts.stride, permissive=opts.permissive)
  File "/home/chem/iorga/temp_ftp/mdpow-pcw", line 99, in run_pcw
    gwat, gcyc = load_gsolv(directory, permissive=kwargs.pop('permissive',False),)
  File "/home/chem/iorga/temp_ftp/mdpow-pcw", line 60, in load_gsolv
    gcyc = mdpow.fep.Gcyc(filename=files['cyclohexane'], basedir=directory, permissive=permissive)
AttributeError: 'module' object has no attribute 'Gcyc'

Add EDR/(Compressed XVG) switch for dH data output

Add an option in the runinput.yml file for the user to specify how they would like energy data to be written out. The options should be between zipped xvg files and an edr file

user yml config overwrites ALL package defaults

Before changing to the yml parser, one could leave values in the user runinput.cfg unset and the parser would use the values from the bundled runinput.cfg. (Technically, this was done by first reading the bundled file and then the user config.)

Now the behavior is to completely replace the package defaults, which also means that the user runinput must be complete or the run will crash somewhere.

I liked the old behavior better but if we decide to use the new one, we must document it.

Opinions?

need tests

In the light of issue #20 we really need tests. That's not easy but we need to get started.

An example is an initial test in mdpow/tests/test_fep.py.

Remove MDrunnerSimple from fep.py

The current implementation of MDrunnerSimple really doesn't add very much to what is provided with gromacs.run.MDrunner.

class MDrunnerSimple(gromacs.run.MDrunner):
    """Gromacs mdrun.

    Use Gromacs 4.5.x with threaded mdrun to get max performance and
    just leave everything as it is. For older version you can change
    the name and add a mpiexec binary to launch a multiprocessor
    job. See :mod:`gromacs.run` for details.

    """
    mdrun = "mdrun"
    mpiexec = None

remove MDrunnerSimple from MDPOW
switch all instantiations of MDrunnerSimple to instantiations of gromacs.run.MDrunner

Use Parrinello-Rahman pressure coupling in MD_NPT

Change the pressure coupling algorithm in the MD_NPT protocol from Berendsen to Parrinello-Rahman. Requires:

Change Pcoupl from Berendsen to Parrinello-Rahman in NPT_opls.mdp
Pass '-t $previous_checkpoint' to GromacsWrapper via MDPOW setupMD function
Run benzene simulations

change config parser to use yml lists instead of getlist()

One advantage of using yml is that it has rich data structures. We should make use of this and change the runinput config syntax to use

deeper nestings (dict-like)
lists

This will also make it easier to implement #2.

However, it will break the input file format so we might want to wait with this until current projects are done...

Examples

Lists

lambdas: 0.0, 0.25, 0.5, 0.75, 1.0

to using YAML lists

lambdas: [0.0, 0.25, 0.5, 0.75, 1.0]

lambdas: 
  - 0.0
  - 0.25
  - 0.5
  - 0.75
  - 1.0

Deeper nesting

Similarly, we should use deeper nestings such as

FEP:
   maxwarn: 1
   mdp: "bar_opls.mdp"
   method: "BAR"
   qscript: "wsq_4_MDPOW_array_default_466.sge"
   runlocal: False
   runtime: 5000
   schedule:
     Coulomb:
           couple_lambda0: "vdw-q"
           couple_lambda1: "vdw"
           description: "dis-charging vdw+q --> vdw"
           label: "Coul"
           lambdas: [0.0, 0.25, 0.5, 0.75, 1.0]
           name: Coulomb
           sc_alpha: 0
           sc_power: 1
           sc_sigma: 0.3
     VDW:
           couple_lambda0: "vdw"
           couple_lambda1: "none"
           description: "decoupling vdw --> none"
           label: "VDW"
           lambdas: [0.0, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.65, 0.7, 0.75, 0.8, 0.85, 0.9, 0.95, 1]
           name: vdw
           sc_alpha: 0.5
           sc_power: 1.0
           sc_sigma: 0.3

mdpow-equilibrium generate template broken

Using the mdpow-equilibrium --get-template script does not return template.

mdpow       : INFO     MDPOW 0.6.1-dev starting.
mdpow       : INFO     Copyright (c) 2010-2016 Ian Kenney, Bogdan Iorga, and Oliver Beckstein
mdpow       : INFO     Released under the GNU Public Licence, version 3.
mdpow       : INFO     For bug reports and help: https://github.com/Becksteinlab/MDPOW/issues
mdpow.config: INFO     Using the bundled force fields from GMXLIB='/nfs/homes/ikenney/Library/python/MDPOW/mdpow/top'.
INFO:mdpow.config:Using the bundled force fields from GMXLIB='/nfs/homes/ikenney/Library/python/MDPOW/mdpow/top'.
mdpow.config: INFO     If required, override this behaviour by setting the environment variable GMXLIB yourself.
INFO:mdpow.config:If required, override this behaviour by setting the environment variable GMXLIB yourself.
mdpow       : CRITICAL A run input file is required See --help.
CRITICAL:mdpow:A run input file is required See --help.

Have paths in test files use os.path.join()

read subject

and comment steps

make FEP schedules configurable on a per solvent basis

Requires #1 to be implemented.

Note that at the moment the default for schedules in Goct are never used because we always pass the schedule kwarg in run.fep_simulation, which is taken from the input cfg file.

use python-quantities instead of numkit.observables

At the moment we use GromacsWrapper's numkit.observables to represent results, mainly to do error propagation. However, there's the mature https://github.com/python-quantities/python-quantities (which even comes with proper unit support) and we should replace the GW solution.

Simulations use absolute paths for file paths

@orbeckst currently the simulation files are using absolute paths to identify structure files. While writing tests (pr #58 ) I have been having trouble dealing with the previously generated files having paths related to where they were created; the tests will fail even though all required files are there. I know of the rebuild simulation script, but is that the only solution to this problem? It seems to me that having the abspaths is just an unnecessary restraint.

prepare a release 0.6.0

check that everything installs cleanly (eg use a virtualenv)
change release to TRUE in __init__.py
make release on GitHub (but only once you're sure that the package works)

mdpow-* script needed for analyzing BAR runs

At the moment there is no script available to calculate free energies from BAR runs. This is essential. If we need to call g_bar then that's ok but it must be wrapped in a mdpow-* script and fit into the standard workflow.

What's a workaround for right now?

test failures due to unknown `-doc` argument

Recently a few tests started failing. Probably because they now use GromacsWrapper 0.6.0 and perhaps something changed there.

Need to check.

...

gromacs.setup: INFO     Solvated system with tip4p.gro
/Users/oliver/.virtualenvs/mdpow/develop/lib/python2.7/site-packages/gromacs/core.py:547: GromacsFailureWarning: Gromacs tool failed
Command invocation: grompp -c solvated.gro -f none.mdp -doc grompp wrapper that only warns on failure but does not raise :exc:`GromacsError` -o topol.tpr -p /private/var/folders/qs/k07z49fh8xl6xd008k8wxhxr0000gp/T/tmpdktKAc/Equilibrium/water/top/system.top -maxwarn 10
Gromacs command 'grompp' fatal error message:
GMX_FATAL  Source code file: /Users/oliver/Biop/Library/Gromacs/code/source/gromacs-4.6.5/src/gmxlib/statutil.c, line: 977
GMX_FATAL
GMX_FATAL  Invalid command line argument:
GMX_FATAL  -doc
GMX_FATAL  For more information and tips for troubleshooting, please check the GROMACS
GMX_FATAL  website at http://www.gromacs.org/Documentation/Errors
Error code: 255

  warnings.warn(msg + '\nError code: %r\n' % rc, category=GromacsFailureWarning)
gromacs.cbook: ERROR    grompp_qtot() failed. See warning and screen output for clues.
=========== grompp (stdout/stderr) ============

...

-------------------------------------------------------
Program grompp, VERSION 4.6.5
Source code file: /Users/oliver/Biop/Library/Gromacs/code/source/gromacs-4.6.5/src/gmxlib/statutil.c, line: 977

Invalid command line argument:
-doc

use YAML for config input file

The ini format used for cfg files is limiting. Switch to yaml.

implement yaml parsing
provide translator script to rewrite old cfg files as yml

Have travis use container based infrastructure

Currently sudo: false cannot work due to the use of apt-get to install libgomp1. Find solution to this in order to use container based infrastructure which will increase testing speed.

deprecate mdpow-ghyd and mdpow-goct

mdpow-solvationenergy from #11 incorporates the functionality of mdpow-ghyd and mdpow-goct and we also need to maintain a single script. Thus we should deprecate the specialized scripts and then remove them for the 0.7 release.

remove scripts from docs
add deprecation warnings and mention that they will be gone for 0.7. Warnings should say how to use mdpow-solvationenergy to accomplish the same goal (i.e. mdpow-solvationenergy --solvent=<solvent>).

mdpow-ghyd fails with NameError

Running mdpow-ghy DIR fails with

Traceback (most recent call last):
  File "/nfs/homes2/oliver/.local/bin/mdpow-ghyd", line 6, in <module>
    exec(compile(open(__file__).read(), __file__, 'exec'))
  File "/nfs/homes2/oliver/Projects/Methods/MDPOW/scripts/mdpow-ghyd", line 224, in <module>
    force=opts.force, stride=opts.stride, permissive=opts.permissive)
  File "/nfs/homes2/oliver/Projects/Methods/MDPOW/scripts/mdpow-ghyd", line 115, in run_ghyd
    gwat.analyze(**akw)
  File "/nfs/homes2/oliver/Projects/Methods/MDPOW/mdpow/fep.py", line 931, in analyze
    DY = numpy.array([x.error[0]  for x in xvgs])   # takes a while: computes correl.time
  File "build/bdist.linux-x86_64/egg/gromacs/fileformats/xvg.py", line 457, in error
  File "build/bdist.linux-x86_64/egg/gromacs/fileformats/xvg.py", line 436, in _correlprop
  File "build/bdist.linux-x86_64/egg/gromacs/fileformats/xvg.py", line 397, in _tcorrel
  File "build/bdist.linux-x86_64/egg/gromacs/analysis/__init__.py", line 77, in <module>
  File "build/bdist.linux-x86_64/egg/gromacs/analysis/plugins/__init__.py", line 147, in <module>
NameError: name 'p' is not defined

This seems to be an issue with GromacsWrapper (I am running the latest develop).

choosing different water models

We want to be able to select the water model as a configuration parameter. For example:

FEP windows should continue from MD_NPT with cpt

At the moment (after #31), MD_NPT runs with Parrinello-Rahman barostat (and Langevin thermostat) but the FEP windows (with the same thermostat and barostat) regenerate velocities. We can consider using continuation runs by passing on checkpoint files with the -t $CPT flag.

Should we do this?

remove output of "Helmholtz" free energy

The code still refers to the calculated free energy a Helmholtz free energy (because we used to work in NVT); since we moved to NPT, we actually directly get the Gibbs free energy.

remove all user-visble output and logging that mentions Helmholtz free energy or rewrite so that it refers to Gibbs (to reduce user confusion...)
update docs
clean up code and variable names

should work with Gromacs 5.x

MDPOW should be using Gromacs 5.1.x, which is now pretty stable. GromacsWrapper recently moved to Gromacs 5.x Becksteinlab/GromacsWrapper#26 so this should work with the latest GW.

default names of output pdf files from mdpow-pow and mdpow-pcw

Currently mdpow-pow and mdpow-pcw scripts generate pdf files with default names dVdl_UNK.pdf and dVdl_UNK_water, respectively, assuming that the ligand code is UNK. I suggest that these names should be dVdl_UNK_pow.pdfand dVdl_UNK_pcw.pdf, respectively.

local test fails due to missing manifest/PyBOL

At the moment I cannot test locally out of the box.

I did a

pip install --upgrade .

but because this does not install PyBOL (manifest) running the tests locally fails. (See below)

make PyBOL pip-installable
add PyBOL to the requirements for tests in setup.py (something like tests_require = [ ... ])

(develop) yngvi:MDPOW oliver$ py.test mdpow/tests/
============================================================= test session starts =============================================================
platform darwin -- Python 2.7.12, pytest-2.9.1, py-1.4.31, pluggy-0.3.1
rootdir: /Volumes/Data/oliver/Biop/Projects/Methods/MDPOW, inifile:
collected 49 items / 2 errors

mdpow/tests/test_emin.py ..
mdpow/tests/test_fep.py ..
mdpow/tests/test_forcefields.py ................................
mdpow/tests/test_runinput.py .........
mdpow/tests/test_solvation.py ....

=================================================================== ERRORS ====================================================================
__________________________________________ ERROR collecting mdpow/tests/test_equilibration_script.py __________________________________________
mdpow/tests/test_equilibration_script.py:3: in <module>
    import manifest
E   ImportError: No module named manifest
_______________________________________________ ERROR collecting mdpow/tests/test_fep_script.py _______________________________________________
mdpow/tests/test_fep_script.py:3: in <module>
    import manifest
E   ImportError: No module named manifest
===================================================== 49 passed, 2 error in 4.55 seconds ======================================================

add pyyaml as dependency

See @iorga's comment on #3. The python yaml package must be declared as a dependency in setup.py.

update tests: py.test 4.0 will not have yield anymore

/home/travis/build/Becksteinlab/MDPOW/mdpow/tests/test_forcefields.py yield tests are deprecated, and scheduled to be removed in pytest 4.0

mdpow-solvation: name default output file according to solvent

The mdpow-solvation script should accept auto keywords for the energy and the plot file and name these files so that they are distinct for the different solvents, eg

plot: dHdl_{solvent}.pdf

At the moment, they use the default for water (ghyd.txt) and always dVdl_{code}.pdf, which overwrites the same files when only --solvent=... is varied.

Minimum distance default leads to instabilities

I have been running some some manual and scripted solvation free energy calculations on benzene in water and have found that the default value of 1nm leads to many failed simulations.

Fatal error: The box size in direction X (2.449481) times the triclinic skew factor (0.816497) is too small for a cut-off of 1.000000 with 2 domain decomposition cells, use 1 or more than 2 cells or increase the box size in this direction

Should we really be setting 1nm to the default?

Improve solvation testing

While the tests currently do what we want, it would be a good idea to check more than just the fact that the solvation process has no errors.

What should be checked:

Paths to solvated structures make sense
The correct solvent is actually used
Multiple solvated structures can coexist in the same 'calculation' filesystem

wrong MDP file for VDW FEP

MDPOW currently creates wrong mdp files for the VDW part of the FEP: The FEP lines read

couple-lambda0 = vdw
couple-lambda1 = vdw-q

but this is wrong for the decoupling step where we switch off LJ.

Instead it must be

couple-lambda0 = vdw
couple-lambda1 = none

(as also indicated in the comments in the file itself!)

Any simulations ran with this buggy version are WRONG. I recommend deleting the whole FEP part and running anew (technically, the Coulomb part is correct but it can get messy figuring out what needs to be rerun so I advocated a clean slate).

manifest_parser name change will break test functionality

manifest_parser name change will break test functionality. Preemptive changes should include:

change all imports (i.e. import manifest --> import pybol )
change all instantiations of Manifest objects (i.e. m = manifest.Manifest() --> m = pybol.Manifest()
change how travis installs pybol

implement MBAR

Set up MDPOW to produce data for MBAR (basically, calculate all windows as foreign lambdas) and then use pymbar for analysis.

Dynamic version calculation breaks clean installation

Dynamic calculation of the version in setup.py circumvents the installation of a dependency (pyyaml) and causes an import error on pip install -e. I removed the offending line and hard coded in the version into setup.py...

@orbeckst do you have any preferences on how I fix this? I see that __init__.py has import config at the bottom (line 540) but is never used. This seems to be the only source of the error. Does this really need to be there?

Commit of hard code: 40cef6a

remove "NPT correction" crud

The correction for NVT to NPT was never properly implemented and all corresponding code removed a while ago. However, there's still some crud in the Gsolv class and the output files left. This should be removed.

output files VdP column: remove (Note: *changes "gsolvation.txt" file format)
logging output
doc strings

variable expansion in config file not working anymore

It used to be possible to define a variable in the DEFAULT section and then use the variable in other sections (the ini parser was able to do it) but with the YAML format this does not work anymore. Thus, the following from the template runinput.yml is not working anymore:

DEFAULT:
    qscripts: local.sh


MD_NPT:
    qscript: (qscripts)s
         # queuing system scripts to produce

add M24 water model for testing

Aim

For testing purposes we should add the M24 water model, introduced by Shirts and Pande¹.

Background

Shirts¹ introduced a range of proof-of-concept TIP3P-like water models. He changed the epsilon and sigma parameters and obtained much better hydration free energies for amino acid sidechain analogs than with any of the regular models. The one termed M24 performed best. These M* models are not meant for production simulations but are proof-of-concept that fixed charge models and in particular water models still have a lot of room in the parametrization.

M24 water model

TIP3P geometry
epsilon: 0.24 kcal/mol
sigma: 3.111 Å

Tasks

OPLS-AA force field

Need to add various entries to the bundled force field files

add entry to watermodels.dat
add atomtype for the oxygen
add new top/oplsaa.ff/m24.itp

In MDPOW

add m24 to mdpow/forcefields.py — wherever tip3p appears one should also have an equivalent m24 entry; note that the corresponding gro file will be spc216.gro.

References

M. R. Shirts and V. S. Pande. Solvation free energies of amino acid side chain analogs for common molecular mechanics water models. J. Chem. Phys., 122(13), 2005. doi:10.1063/1.1877132

one-shot do-all workflow script

We want a solution to run the whole protocol (or workflow) with a single command.

Possible approaches

daemon/nanny process
chains of Gridengine jobs
...

coverage should not contain the tests

PR #39 adds testing with Gromacs 4 and Gromacs 5 (yay!). However, coverage also analyses coverage of the tests themselves. That's not helpful. All tests ought to be excluded from coverage reporting.

There should be some simple settings for coverage in the travis.yml file that exclude tests.

MDP errors with gromacs 5.X.X

In gromacs 4.X.X mdps using a non-integer value for sc-power could generate a tpr. This is not the case with gromacs 5.X.X. During window setup, a float value is substituted in at some level. This needs to be an integer.

ERROR 1 [file md.mdp, line 140]:
  Right hand side '1.0' for parameter 'sc-power' in parameter file is not
  an integer value



Back Off! I just backed up mdout.mdp to ./#mdout.mdp.1#

There was 1 note

-------------------------------------------------------
Program gmx grompp, VERSION 5.1.3
Source code file: /nfs/homes4/dldotson/Sysadmin/install/gromacs/source/gromacs-5.1.3/src/gromacs/gmxlib/readinp.c, line: 287

Fatal error:
There was 1 error in input file(s)
For more information and tips for troubleshooting, please check the GROMACS
website at http://www.gromacs.org/Documentation/Errors
-------------------------------------------------------

set up continuous integration on Travis CI

With tests #23 we also want to do CI. See eg MDAnalysis for how to set it up.

Packaging files for use in tests

Need an efficient way of packaging states of an MDPOW workflow to be used as intermediate points in testing.

For instance, if testing the the MD_NPT protocol, one would need to start just after the emin protocol. However, we need to keep all tests independent of each other; using the files from previous tests break that rule. Is the following setup really the best:

+ --- /testing_resources/states
       + --- Emin
               + --- Solvation
       + --- MD_NPT
               + --- Solvation --> Emin 
       + --- FEP
               + --- Solvation --> Emin --> MD_NPT
       + --- Analysis
               + --- Solvation --> Emin --> MD_NPT --> FEP

This also raises the question involving trajectories that will be used for analysis. Do we just use shorter trajectories?

Ligand code containing only digits and starting with 0 incorrectly recognized as number

When the ligand code contains only digits and starts with 0 (001 in the following example), it is incorrectly recognized as number and the subsequent steps fail:

> mdpow-equilibrium --solvent=water resources/runinput.yml
mdpow       : INFO     MDPOW 0.6.1-dev starting.
mdpow       : INFO     Copyright (c) 2010-2016 Ian Kenney, Bogdan Iorga, and Oliver Beckstein
mdpow       : INFO     Released under the GNU Public Licence, version 3.
mdpow       : INFO     For bug reports and help: https://github.com/Becksteinlab/MDPOW/issues
NOTE: Some configuration directories are not set up yet
      ['/home/chem/iorga/.gromacswrapper/managers']
      You can create them with the command from within Python
      >>> import gromacs
      >>> gromacs.config.setup()
mdpow.config: INFO     Using the bundled force fields from GMXLIB='/sge/packs_centos/opt/python-2.7.10-CentOS6/lib/python2.7/site-packages/mdpow/top'.
INFO:mdpow.config:Using the bundled force fields from GMXLIB='/sge/packs_centos/opt/python-2.7.10-CentOS6/lib/python2.7/site-packages/mdpow/top'.
mdpow.config: INFO     If required, override this behaviour by setting the environment variable GMXLIB yourself.
INFO:mdpow.config:If required, override this behaviour by setting the environment variable GMXLIB yourself.
DEBUG:mdpow.config:Loaded runinput defaults from '/sge/packs_centos/opt/python-2.7.10-CentOS6/lib/python2.7/site-packages/mdpow/templates/runinput.yml'
DEBUG:mdpow.config:Loaded user runinput from 'resources/runinput.yml'
mdpow.config: WARNING  Changing GromacsWrapper environment: flags['capture_output'] = True
WARNING:mdpow.config:Changing GromacsWrapper environment: flags['capture_output'] = True
DEBUG:mdpow.run:MD_NPT: Using MDP file 'NPT_opls.mdp' from config file
DEBUG:mdpow.run:MD_relaxed: Using MDP file 'NPT_opls.mdp' from config file
DEBUG:mdpow.run:energy_minimize: Using MDP file 'em_opls.mdp' from config file
mdpow.run   : INFO     Selected water model: m24
INFO:mdpow.run:Selected water model: m24
DEBUG:gromacs.utilities:Working in '/gem/iorga/projects/sampl4_sims_NPT_m24/SAMPL4_001_1/SAMPL4_001_1/Equilibrium/water/top'...
gromacs.cbook: INFO     editing txt = '/sge/packs/packs_centos/opt/python-2.7.10-CentOS6/lib/python2.7/site-packages/mdpow/templates/system.top' (5 substitutions)
INFO:gromacs.cbook:editing txt = '/sge/packs/packs_centos/opt/python-2.7.10-CentOS6/lib/python2.7/site-packages/mdpow/templates/system.top' (5 substitutions)
DEBUG:gromacs.cbook:match:    #include "compound.itp"
DEBUG:gromacs.cbook:replaced: #include "SAMPL4_001_1.itp"
DEBUG:gromacs.cbook:match:    #include "oplsaa.ff/tip4p.itp"
DEBUG:gromacs.cbook:replaced: #include "oplsaa.ff/m24.itp"
DEBUG:gromacs.cbook:match:    Compound DRUG in solvent (OPLS-AA parameters)
DEBUG:gromacs.cbook:replaced: Compound DRUG in water (OPLS-AA parameters)
DEBUG:gromacs.cbook:match:    Compound DRUG in water (OPLS-AA parameters)
DEBUG:gromacs.cbook:replaced: Compound 1 in water (OPLS-AA parameters)
DEBUG:gromacs.cbook:match:    DRUG               1
DEBUG:gromacs.cbook:replaced: 1               1
gromacs.cbook: INFO     edited txt = 'system.top'
INFO:gromacs.cbook:edited txt = 'system.top'
mdpow.equil : INFO     [/gem/iorga/projects/sampl4_sims_NPT_m24/SAMPL4_001_1/SAMPL4_001_1/Equilibrium/water/top] Created topology 'system.top' that includes 'SAMPL4_001_1.itp'
INFO:mdpow.equil:[/gem/iorga/projects/sampl4_sims_NPT_m24/SAMPL4_001_1/SAMPL4_001_1/Equilibrium/water/top] Created topology 'system.top' that includes 'SAMPL4_001_1.itp'
DEBUG:gromacs.cbook:Scrubbed [ molecules ]: marker = '; Gromacs auto-generated entries follow:'
DEBUG:gromacs.utilities:Working in '/gem/iorga/projects/sampl4_sims_NPT_m24/SAMPL4_001_1/SAMPL4_001_1/Equilibrium/water/solvation'...
gromacs.setup: INFO     [/gem/iorga/projects/sampl4_sims_NPT_m24/SAMPL4_001_1/SAMPL4_001_1/Equilibrium/water/solvation] Solvating with water 'spc216.gro'...
INFO:gromacs.setup:[/gem/iorga/projects/sampl4_sims_NPT_m24/SAMPL4_001_1/SAMPL4_001_1/Equilibrium/water/solvation] Solvating with water 'spc216.gro'...
DEBUG:gromacs.core:editconf -f /gem/iorga/projects/sampl4_sims_NPT_m24/SAMPL4_001_1/resources/SAMPL4_001_1.pdb -o boxed.gro -bt dodecahedron -d 1.0
DEBUG:gromacs.core:genbox -cs spc216.gro -p /gem/iorga/projects/sampl4_sims_NPT_m24/SAMPL4_001_1/SAMPL4_001_1/Equilibrium/water/top/system.top -cp boxed.gro -o solvated.gro
gromacs.setup: INFO     Solvated system with spc216.gro
INFO:gromacs.setup:Solvated system with spc216.gro
DEBUG:gromacs.core:grompp -p /gem/iorga/projects/sampl4_sims_NPT_m24/SAMPL4_001_1/SAMPL4_001_1/Equilibrium/water/top/system.top -c solvated.gro -maxwarn 1 -o topol.tpr -f none.mdp
/sge/packs_centos/opt/python-2.7.10-CentOS6/lib/python2.7/site-packages/gromacs/core.py:544: GromacsFailureWarning: Gromacs tool failed
Command invocation: grompp -p /gem/iorga/projects/sampl4_sims_NPT_m24/SAMPL4_001_1/SAMPL4_001_1/Equilibrium/water/top/system.top -c solvated.gro -maxwarn 1 -o topol.tpr -f none.mdp
Gromacs command 'grompp' fatal error message:
GMX_FATAL  Source code file: /tmp/gromacs-4.6/src/kernel/toppush.c, line: 2316
GMX_FATAL  
GMX_FATAL  Fatal error:
GMX_FATAL  No such moleculetype 1
GMX_FATAL  For more information and tips for troubleshooting, please check the GROMACS
GMX_FATAL  website at http://www.gromacs.org/Documentation/Errors
Error code: 255
...

The content of the .yml file is the following:

> cat resources/runinput.yml 
DEFAULT:
FEP:
  maxwarn: 2
  mdp: bar_opls.mdp
  method: BAR
  qscript: cluster8.sge
  runlocal: False
  runtime: 5000
FEP_schedule_Coulomb:
  couple_lambda0: vdw-q
  couple_lambda1: vdw
  description: dis-charging vdw+q --> vdw
  label: Coul
  lambdas: 0.0, 0.25, 0.5, 0.75, 1.0
  name: Coulomb
  sc_alpha: 0
  sc_power: 1
  sc_sigma: 0.3
FEP_schedule_VDW:
  couple_lambda0: vdw
  couple_lambda1: none
  description: decoupling vdw --> none
  label: VDW
  lambdas: 0.0, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.65, 0.7, 0.75, 0.8, 0.85, 0.9,
    0.95, 1
  name: vdw
  sc_alpha: 0.5
  sc_power: 1.0
  sc_sigma: 0.3
MD_NPT:
  maxwarn: 2
  mdp: NPT_opls.mdp
  qscript: cluster8.sge
  runlocal: False
  runtime: 15000
MD_relaxed:
  maxwarn: 2
  mdp: NPT_opls.mdp
  qscript: local8-cluster.sh
  runlocal: True
  runtime: 5
energy_minimize:
  mdp: em_opls.mdp
mdrun:
  maxthreads: None
  nice: 19
  stepout: 100
  verbose: True
setup:
  gromacsoutput: False
  itp: resources/SAMPL4_001_1.itp
  maxwarn: 1
  watermodel: m24
  molecule: 001
  name: SAMPL4_001_1
  structure: resources/SAMPL4_001_1.pdb

Use NPT by default for FEP

Switch mdp files for fep to use Parrinello-Rahman p coupling.