Immunogenicity B-Cell Epiptope¶

Predicts B-cell epitope immunogenicity for one or more protein structures using a sliding window across the sequence. It summarizes predicted epitope propensities into a CSV table for downstream analysis. Supports production evaluation, a fast test mode, and a mock mode with precomputed data.

Usage¶

Use this node to assess potential B-cell epitope regions on designed or candidate protein structures. Typical workflow: generate or load PDB structures, run this node to scan for epitope propensity with a chosen window size, and use the resulting CSV to compare immunogenicity across designs or to guide sequence/structure refinement.

Inputs¶

Field	Required	Type	Description	Example
pdb	True	PDB	One or more protein structures to evaluate. Provide a mapping from a unique name to the PDB content.	{"design_1": "ATOM ...\nEND", "design_2": "ATOM ...\nEND"}
window_size	True	INT	Sliding window length used to scan the sequence for B-cell epitope propensities. Larger windows smooth local variation; smaller windows give finer granularity.	10
mode	True	["MOCK", "PROD", "TEST"]	Execution mode. PROD runs the full evaluation, MOCK returns precomputed example results, TEST runs a faster, minimal evaluation (may limit inputs).	PROD

Outputs¶

Field	Type	Description	Example
score.csv	CSV	CSV table of B-cell epitope predictions for the provided proteins. Contains rows for each protein (and sliding window region) with corresponding scores.	protein,window_start,window_end,epitope_score\ndesign_1,1,10,0.42\ndesign_1,2,11,0.37

Important Notes¶

The window_size strongly affects sensitivity and smoothness of predictions; adjust based on protein length and desired resolution.
In TEST mode, the node may limit evaluation to the first provided structure to speed up runtime.
MOCK mode returns precomputed example results and is intended for UI testing and demonstrations, not scientific conclusions.
If multiple structures are provided, the output CSV will contain aggregated rows for all inputs.
Ensure input PDBs are valid protein structures; malformed or non-protein content can lead to failed or meaningless predictions.

Troubleshooting¶

No or empty results: Verify the pdb input contains at least one valid structure and that window_size is a positive integer.
Unexpectedly few rows: In TEST mode, only the first structure may be evaluated. Switch to PROD mode or reduce the number of inputs.
CSV appears malformed: Check for special characters or commas in protein names. Use simple alphanumeric identifiers for pdb keys.
Long runtimes: Reduce the number of input structures, use a larger window_size, or switch to TEST mode to validate configuration before full runs.