Overview ======== DROPPS (*Distributed Rapid Operation Platform for Phase-separation Simulations*) is a unified, GROMACS-style workflow platform for performing large-scale coarse-grained (CG) simulations of biomolecular liquid–liquid phase separation (LLPS). The system was designed to solve long-standing issues in the LLPS simulation community, including workflow fragmentation, force-field inconsistencies, and a lack of standardized tooling for multichain CG modeling. DROPPS integrates **OpenMM**, **MDAnalysis**, custom high-efficiency IO layers, a modular force-field engine, and a GROMACS-inspired command-line interface (“``dps``”). Its goal is to make CG LLPS simulations as easy and reproducible as running a typical GROMACS simulation — but with significantly greater flexibility and model extensibility. Why DROPPS Exists ----------------- Traditional CG simulations for LLPS often involve: - many force fields scattered across many folders - per-project ad-hoc scripts - duplicated coarse-graining procedures - incompatible file formats - fragile user-built analysis pipelines - impossible reproducibility across research groups DROPPS solves these issues by: - providing **one** unified CLI - integrating **all** CG workflows into a single platform - offering a stable I/O and FF engine - providing an extendable command registry - offering a Python API for analysis, system building, and FF manipulation Design Philosophy ----------------- DROPPS is built around five core principles (pages 6–8): - **统一 (Unification)** — all workflows follow the same logical structure. - **通用 (General-purpose)** — supports any CG model, not only HPS. - **兼容 (GROMACS-like interface)** — familiar commands for experienced users. - **易用 (Usability)** — reduces the complexity of setting up LLPS simulations. - **可扩展 (Extensibility)** — new FFs, new commands, and new analysis modules can be added with minimal effort. The framework is intentionally non-monolithic. While GROMACS includes its own MD engine, DROPPS *outsources* the expensive parts of the computation to: - **OpenMM** → MD engine - **MDAnalysis** → trajectory handling - **MDTraj / custom readers** → optional readers Architecture ------------ DROPPS consists of four major layers: 1. **Command Layer (CMD)** All user-facing commands such as ``pdb2dps``, ``genmesh``, ``grompp``, ``mdrun``, ``contact``, ``density``, etc. Registered in ``all_commands``. 2. **pydps Layer** The core Python package implementing structure generation, FF processing, topology handling, mesh operations, MD input conversion, and analysis routines. 3. **IO & Utilities Layer (fileio / misc)** - Topology reader/writer - PDB/ITP/TOP parsers - MDP parser - Random generators - Logging and text formatting (rich) - High-performance table utilities 4. **Trajectory Layer (OpenMM + MDAnalysis + custom wrappers)** A unified trajectory object exposing: - OpenMM topology - MDAnalysis Universe - consistent IDs / chain indices / residue labels - helper methods for fast selection and iteration These layers produce a modular yet coherent system. GROMACS-Style Simulation Workflow --------------------------------- DROPPS intentionally mirrors the familiar GROMACS workflow (page 34): =================== =================== GROMACS DROPPS =================== =================== pdb2gmx pdb2dps editconf editconf insert-molecules genmesh grompp grompp mdrun mdrun make_ndx make_ndx analysis tools density / contact / cmap / idist / odist / gyrate =================== =================== This ensures that experienced GROMACS users can adopt DROPPS with near-zero learning cost. Key Features ------------ DROPPS provides the following capabilities: **Unified workflow** A full simulation pipeline that covers: - PDB/sequence coarse-graining - topology assembly - system construction (multi-chain mesh packing) - MD input generation - MD execution - trajectory analysis - visualization utilities **Flexible force-field system** Full programmatic access to: - masses, charges, λ parameters - σ and interaction tables - bonded / nonbonded terms - custom FF loading **High-performance trajectory handling** The custom trajectory wrapper integrates: - OpenMM topology - MDAnalysis coordinates - chain/residue indexing - per-chain, per-group iteration **Extensible command registry** Adding a command requires only: 1. argument parser 2. processing function 3. registering via ``single_command`` **Professional data analysis suite** Provided analysis commands: - ``density`` - ``gyrate`` - ``cmap`` - ``idist`` / ``odist`` - ``contact`` - ``angle`` - ``check`` (topology/trajectory validator) **User mode vs developer mode** DROPPS exposes two “modes” of operation: - **User mode** — simplified commands, preconfigured workflows - **Developer mode** — direct access to topology objects, FF tables, trajectory API, etc. Command Dispatcher System ------------------------- One of the novel design elements of DROPPS is the command-dispatcher architecture: .. code-block:: python all_commands = { "pdb2dps": pdb2dps_cmd, "grompp": grompp_cmd, "mdrun": mdrun_cmd, ... } Users may extend or replace built-in commands by editing ``all_commands`` or by plugging in custom modules. File Formats ------------ DROPPS uses file formats analogous to the GROMACS ecosystem: - **.pdb** — CG coordinate files - **.itp** — single-chain topologies - **.top** — combined topology after system construction - **.mdp** — MD parameter files - **.tpr** — DROPPS-specific MD input (not compatible with GROMACS) - **.xtc** — compressed trajectory files - **.ndx** — selection groups Important note: DROPPS `.tpr` and GROMACS `.tpr` are structurally incompatible. They cannot be interconverted. Target Users ------------ DROPPS is designed for: - LLPS researchers - protein-condensate modelers - large-system CG simulation users - method developers who need to integrate custom FFs - students who need a standardized simulation platform - PIs who want reproducible workflows across group members The platform’s goal is to replace the per-student, per-project “script zoo” with a unified and professional tool. Summary ------- DROPPS provides: - a unified, GROMACS-inspired interface - robust force-field and topology processing - an extensible MD pipeline using OpenMM - high-performance trajectory tools - a rich and extendable analysis suite The entire ecosystem is structured to make CG LLPS simulations **repeatable, maintainable, and shareable**, with minimal user-side scripting and maximal developer flexibility.