README.md

Gromacs and Orca Kubernetes wrapper

Every minor software package needed is contained in the main pipeline image. Gromacs and Orca are considered too large to be contained in the main container. Therefore these packages are divided into another two independent images.

This implies there is a need of connecting the main pipeline container with Gromacs and Orca containers. A wrapper has been written exactly for this purpose. The wrapper is written in Python and converts the Gromacs/Orca command along with its arguments to a .yaml (Kubernetes job) configuration file. Based on this file the Kubernetes job is spawned where all computations happen.

Gromacs

gmx_run()

Mandatory argument:

string gmx_command       -       gromacs command to be evaluated

Optional arguments:

int mpi_run              -       request number of cpus for mpi_run software
string groups            -       specify interactive groups for gromacs (1)
string make_ndx          -       specify interactive make_ndx input for gromacs make_ndx (2)
string image             -       specify used gromacs image
string workdir           -       specify directory where should the calculation take place
string arch              -       specify CPU architecture
bool double              -       double precision for mdrun (3)
bool rdtscp              -       enable rdtscp
bool parallel            -       run jobs in parallel (independently). Disabled by default

1. https://manual.gromacs.org/documentation/2019-rc1/reference-manual/analysis/using-groups.html
2. https://manual.gromacs.org/archive/5.0.4/programs/gmx-make_ndx.html
3. https://manual.gromacs.org/current/onlinehelp/gmx-mdrun.html

---

Example usages:

gmx_run('mdrun -deffnm em1 -ntomp 2')
gmx_run('mdrun -deffnm mtd1 -replex 500 -plumed plumed.dat', mpi_run=6')
gmx_run('mdrun -deffnm em1', image='user/gromacs:123')
gmx_run('mdrun -deffnm em1 -ntomp 2', arch='AVX256')
gmx_run('mdrun -deffnm em1 -ntomp 2', workdir='/home', double=True, arch='AVX256')

Orca

orca_run()

Mandatory arguments:

string orca_command      -       orca method (file) used for computation (1)
string log               -       log file to store the output of computation

Optional arguments:

string image             -       specify used Orca image
string workdir           -       specify directory where should the calculation take place
bool parallel            -       run jobs in parallel (independently). Disabled by default

1. https://sites.google.com/site/orcainputlibrary/generalinput

---

Example usages:

orca_run('method1.inp', 'output1.out", workdir='/home', parallel=True)

Gromacs/Orca

parallel_wait()

This method has to be placed after a gmx_run() or orca_run() with parallel enabled. It is a wait mechanism to detect if all jobs running in parallel finished. Afterwards logs of these jobs are printed.

orca_run('method1.inp', 'output1.out", workdir='/home', parallel=True)
orca_run('method2.inp', 'output2.out", workdir='/home', parallel=True)

parallel_wait()

---

for i in range 6:
    orca_run(f'method{i}.inp', f'output{i}.out", workdir='/home', parallel=True)
parallel_wait()