Overview

SODAPOP (Structure-based Open Docking and Assessment Pipeline fOr Protein assemblies) is a webserver that provides an end-to-end pipeline with programs for protein-protein docking (LightDock), decoy scoring (DS2), and interface evaluation (PPCheck).

Submission Guide

Docking Simulations

Available on the Docking submission tab, LightDock allows for ab initio blind docking between pairs of partners.

• Using Local Files: Upload the individual unbound .pdb files for both your Receptor and Ligand.
• Using PDB IDs: Enter valid 4-character RCSB PDB identifiers (e.g., 1AK4 and 1Y64). The server handles fetching and parsing seamlessly.

Scoring Framework

Through the Scoring submission tab, users can upload batch archives to evaluate decoys using either DockScore 2.0 or PPCheck:

• DockScore 2.0: Requires the unbound receptor and ligand .pdb structures directly alongside a compressed .zip file containing the pool of decoy models.
• PPCheck: Upload a .zip archive of decoy complexes. The system unzips and cleans formatting boundaries, and executes the PPCheck pipeline on all files concurrently.

Parameters & Defaults

Default Execution Parameters

To ensure server stability and equitable resource distribution, SODAPOP restricts certain intensive simulation parameters. The following default LightDock limits apply to your jobs:

• Number of Swarms: 10 (number of spatial discretization clusters over the receptor surface)
• Number of Glowworms: 10 (number of glowworms inside each swarm)
• Simulation Steps: 50 (number of iterations LightDock runs during the search/sampling phase)

LightDock requirements on this server

Note: All lines other than ATOM records (e.g. HETATM, REMARK, CONECT) are removed from your input files before processing. This is required for compatibility with the DFIRE scoring function used by LightDock.

DockScore 2.0 requirements

DockScore 2.0 requires strict atomistic and chain identification correspondence between the unbound receptor/ligand structures mapping to generated decoy subsets. Errors will propagate if disjoint residues exist between the unbound structures and the zipped decoys.

Results & Data Retention

Job Queuing & Monitoring

Heavy jobs are placed in a background queue. An automatically generated unique URL tracks your status and lets you fetch updates. Live terminal outputs stream directly to your browser.

Retrieving Results

Process completion yields an interactive visualizer block. LightDock produces representative structures rendered in real-time WebGL, whereas scoring algorithms export interactively sortable and downloadable tables.

For LightDock jobs, the top 20 decoys are available for individual download and can be submitted directly to DS2 or PPCheck for scoring from the job results page or 3D viewer. The 3D viewer displays the top 10 simultaneously for browser performance; all 20 are included in the bulk .zip download and any scoring submission.

7-Day Retention Policy

Frequently Asked Questions

Why does my job remain in the 'PENDING' state?

This occurs when prior jobs occupy execution cores. Pending jobs patiently wait in the background orchestrator and launch securely when free workers become available.

What does a 'FAILED' status indicate?

Failures are almost universally traced to anomalous .pdb constraints—such as unrecognized heteroatoms, ambiguous missing coordinate headers, or faulty chain indexing. Inspect the integrated Live Log block for specific parser breakpoints.

Can I close the browser tab during a long run?

Yes. The generated results endpoint URL acts as your persistent tracking receipt. Bookmark this link and retrieve the results whenever the cycle finishes.

Why do the DockScore 2.0 and PPCheck results not correlate/match?

DockScore 2.0 and PPCheck are used for different purposes here, and provide orthogonal metrics. DockScore 2.0 is a learned scoring function that ranks poses based on features predictive of native-like interactions. PPCheck evaluates physical interactions using pseudoenergies, and calculates the total stabilizing energy at a given interface. Consequently, they may rank the same poses differently based on their respective scoring functions.