Workflows

The repository contains Nextflow scripts under workflows/. The two-stage flow is built around:

• workflows/build.nf — generate system folders and build files from CSV
• workflows/simulate.nf — run the GROMACS simulation chain

Those scripts match the current mdfactory prepare-build and mdfactory build commands plus the GROMACS file layout produced by the builders.

Full pipeline

The easiest way to run both stages is the provided shell script:

cd workflows/
bash run_full_pipeline.sh

Edit the variables at the top of run_full_pipeline.sh to set your CSV input and output directory. The script runs the build stage followed by the simulation stage.

Stage 1: generate build folders

build.nf expects a CSV input file and an output directory:

nextflow run workflows/build.nf \
  --csv_file systems.csv \
  --output_dir results

What it does:

1. Runs mdfactory prepare-build.
2. Reads the summary YAML written into the output directory.
3. Builds each generated hash directory with mdfactory build.

The summary YAML filename is derived from the CSV stem. For example, systems.csv produces results/systems.yaml.

Stage 2: run the GROMACS chain

simulate.nf consumes the stage-1 output directory plus the summary YAML:

nextflow run workflows/simulate.nf \
  -c workflows/simulate.config \
  --base_dir results \
  --config_yaml results/systems.yaml

simulate.nf executes this fixed sequence for each hash listed in the summary YAML:

1. Minimization using em.mdp
2. NVT equilibration using nvt.mdp
3. NPT equilibration using npt.mdp
4. Production using md.mdp

Outputs are published back into each simulation directory under --base_dir.

Expected directory shape

Next steps