Molclaw Prolif Protein Protein

ProLIF Protein-Protein Interface Fingerprinting Skill

Note:

• Local files are not directly accessible by the server. Please upload them to the server using molclaw-file-transfer before execution.
• For PDB file inputs, it is recommended to preprocess them using molclaw-pdbfixer before execution.
• Please refer to skill molclaw-scp-server to complete tool invocation.

> [!NOTE] > Local files are not directly accessible by the server. Please upload them to the server using molclaw-file-transfer before execution. > For PDB file inputs, it is recommended to preprocess them using molclaw-pdbfixer before execution.

Task Description

Analyze protein-protein interaction trajectories and generate interface interaction fingerprints. Use this skill to evaluate interface stability and identify key residue contributions across simulation.

> Routing note: This tool is the primary choice for protein-protein trajectory interface profiling (multi-frame analysis). For single-structure protein-protein interface analysis, use molclaw-interaction-visualizer in protein mode instead — it produces interface heatmaps, network diagrams, and decision-ready JSON.

Input Source Mapping

| Parameter | Source Guidance | |-----------|-----------------| | topology_path | System topology from MD tools: e.g., protein_openmm_md, prepare_protein_md, goca_pipeline | | trajectory_path | Trajectory from the same MD tools, containing dynamic information for both protein chains | | selection_a | User-defined selection string for protein chain A, for example segid A or protein and chainid A | | selection_b | User-defined selection string for protein chain B, for example segid B or protein and chainid B |

Usage

Tool: prolif_protein_protein

Analyze a protein-protein trajectory and return interaction fingerprints or counts with summary metrics.
Args:
    topology_path (str): Path to the system topology file.
    trajectory_path (str): Path to the trajectory file.
    selection_a (str): Selection string for partner A.
    selection_b (str): Selection string for partner B.
    interactions (List[str]|None): Optional interaction types to compute.
    count (bool): If True, compute interaction counts instead of fingerprints. Default: False.
    vicinity_cutoff (float|None): Optional distance cutoff for vicinity interactions.
    params_json (str|None): Optional JSON parameter file path for ProLIF interaction settings.
    start (int|None): Optional start frame index.
    stop (int|None): Optional stop frame index (exclusive).
    step (int|None): Optional frame stride.
Return:
    status (str): 'success' or 'error'.
    msg (str): Human-readable summary or error message.
    command (str): The executed command label ('protein-protein').
    output_dir (str|None): Run-specific directory under tool_result/prolif_result.
    output_file (str|None): Path to the generated CSV file.
    n_frames (int|None): Number of processed frames.
    n_interactions (int|None): Number of interaction columns in output.
    frequent_interactions (List[dict]|None): High-frequency interactions (>30%) with keys 'interaction' and 'frequency'.
    result_summary (dict|None): Full summary dictionary from the wrapper.

How To Use prolif_protein_protein

response = await client.session.call_tool(
    "prolif_protein_protein",
    arguments={
        "topology_path": "relative/path/to/system.prmtop",
        "trajectory_path": "relative/path/to/md.nc",
        "selection_a": "segid A",
        "selection_b": "segid B",
        "start": 0,
        "step": 10
    }
)
result = client.parse_result(response)
key_output = result["output_file"]

Example Parameter Sets

# 1) Main mode
{
    "topology_path": "relative/path/to/system.prmtop",
    "trajectory_path": "relative/path/to/md.nc",
    "selection_a": "segid A",
    "selection_b": "segid B",
    "start": 0,
    "step": 10
}

# 2) Variant mode
{
    "topology_path": "relative/path/to/system.prmtop",
    "trajectory_path": "relative/path/to/md.nc",
    "selection_a": "protein and chainid A",
    "selection_b": "protein and chainid B",
    "count": True,
    "vicinity_cutoff": 3.5,
    "stop": 200
}

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: InternScience
• Source: InternScience/MolClaw
• License: MIT

Install and usage instructions live in the source repository linked above.