RF Diffusion Contigmap Config¶
Builds a contig map configuration for RF Diffusion protein design runs. It parses user-friendly string ranges into a structured configuration, handling sequence/structure inpainting, provided (unmasked) sequence regions, and overall length constraints. Empty inputs become nulls, and non-length fields are converted to lists of segments.
Usage¶
Use this node to prepare the contig mapping parameters for an RF Diffusion generation workflow. Connect its output to the RF Diffusion node's contigmap_config input to control which residues are masked/unmasked in sequence or structure and to optionally constrain target length.
Inputs¶
| Field | Required | Type | Description | Example |
|---|---|---|---|---|
| inpaint_seq | True | STRING | Residue segments whose sequence should be masked (i.e., to be inpainted). Accepts comma-separated contigs syntax; typically chain-lettered ranges or positions. | A100-120,B50-60 |
| inpaint_str | True | STRING | Residue indices whose 3D structure should be masked (structure inpainting). Comma-separated ranges. Often chain-prefixed. | B165-178 |
| provide_seq | True | STRING | Sequence regions to unmask/provide explicitly (kept fixed). Comma-separated inclusive ranges. | 172-177,200-205 |
| length | True | STRING | Target sequence length or a length range. When set, used as a string constraint rather than a list. | 100-150 |
Outputs¶
| Field | Type | Description | Example |
|---|---|---|---|
| contigmap_config | JSON | Configuration dictionary for RF Diffusion contig mapping with keys: inpaint_seq (list or null), inpaint_str (list or null), provide_seq (list or null), length (string or null). | {'inpaint_seq': ['A100-120', 'B50-60'], 'inpaint_str': ['B165-178'], 'provide_seq': ['172-177', '200-205'], 'length': '100-150'} |
Important Notes¶
- Inputs are parsed as comma-separated lists; surrounding brackets or quotes are stripped automatically.
- Empty strings become null in the output; non-length fields become lists of segments, while length remains a string.
- Ranges are inclusive. Use optional chain identifiers where appropriate (e.g., A100-120).
- This node only prepares configuration; it does not validate biological plausibility or enforce chain existence.
Troubleshooting¶
- Unexpected nulls in output: Ensure the input fields are not left empty or filled with only spaces.
- Downstream node rejects contig segments: Verify your segment format (e.g., use commas to separate, ranges like 100-120, and chain labels where required).
- Length not respected downstream: Confirm that the RF Diffusion node is in a mode that uses length constraints and that length is provided as a single value or 'min-max' string.
- Inconsistent masking behavior: Double-check that sequence masking (inpaint_seq) and structure masking (inpaint_str) target the intended ranges and chains.
Example Pipelines¶
