File size: 4,193 Bytes
373e2bb | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 | # Inference JSON Format
OpenDDE input is a JSON file whose top-level value is a non-empty list of jobs.
It uses AlphaFold Server-style entity keys (`proteinChain`, `dnaSequence`,
`rnaSequence`, `ligand`, `ion`), not the single-job `alphafold3` dialect.
Minimal job:
```json
[
{
"name": "example_job",
"modelSeeds": [101],
"sequences": [
{
"proteinChain": {
"sequence": "ACDEFGHIKLMNPQRSTVWY",
"count": 1
}
}
]
}
]
```
`covalent_bonds` is optional and is omitted here; see the section below for when
to add it.
Job fields:
| Field | Required | Meaning |
| --- | :---: | --- |
| `name` | Yes | Job name used in output paths. |
| `sequences` | Yes | List of entities. Each item has exactly one entity key. |
| `modelSeeds` | No | Default seeds for the job. Overridden by `--seeds`; if neither is set, a random seed is sampled. |
| `covalent_bonds` | No | Explicit covalent links between entities. |
Every entity has `count`. Optional `id` is a list of chain IDs; its length must
match `count`.
## `proteinChain`
```json
{
"proteinChain": {
"sequence": "ACDEFGHIKLMNPQRSTVWY",
"count": 1,
"id": ["A"],
"modifications": [
{"ptmType": "CCD_MSE", "ptmPosition": 1}
],
"pairedMsaPath": "/absolute/path/to/pairing.a3m",
"unpairedMsaPath": "/absolute/path/to/non_pairing.a3m",
"templatesPath": "/absolute/path/to/hmmsearch.a3m"
}
}
```
- `sequence`: 20 standard amino-acid letters plus `X`.
- `ptmType`: CCD code prefixed with `CCD_`; `ptmPosition` is 1-based.
- `pairedMsaPath`, `unpairedMsaPath`: optional protein A3M files.
- `templatesPath`: optional template hits file (`.a3m` or `.hhr`), used only with
`--use_template true`.
## `dnaSequence`
```json
{
"dnaSequence": {
"sequence": "GATTACA",
"count": 1,
"id": ["D"],
"modifications": [
{"modificationType": "CCD_6MA", "basePosition": 2}
]
}
}
```
- Supported documented letters: `A`, `T`, `G`, `C`, `N`, `X`.
- DNA is single-stranded; add another `dnaSequence` for the other strand.
- `basePosition` is 1-based.
## `rnaSequence`
```json
{
"rnaSequence": {
"sequence": "GUAC",
"count": 1,
"id": ["R"],
"modifications": [
{"modificationType": "CCD_5MC", "basePosition": 4}
],
"unpairedMsaPath": "/absolute/path/to/rna_msa.a3m"
}
}
```
- Supported documented letters: `A`, `U`, `G`, `C`, `N`, `X`.
- `unpairedMsaPath` is optional and used only with `--use_rna_msa true`.
## `ligand`
```json
{
"ligand": {
"ligand": "CCD_ATP",
"count": 1,
"id": ["L"]
}
}
```
`ligand` can be:
- A CCD code prefixed with `CCD_`, e.g. `CCD_ATP`.
- Multiple CCD codes joined by underscores, e.g. `CCD_NAG_BMA_BGC`.
- A 3D ligand file prefixed with `FILE_` (`.pdb`, `.sdf`, `.mol`, `.mol2`).
- A SMILES string.
## `ion`
```json
{
"ion": {
"ion": "MG",
"count": 2,
"id": ["M", "N"]
}
}
```
Ion codes are CCD component names without the `CCD_` prefix.
## `covalent_bonds`
```json
"covalent_bonds": [
{
"entity1": "1",
"copy1": 1,
"position1": "2",
"atom1": "SG",
"entity2": "2",
"copy2": 1,
"position2": "1",
"atom2": "C1"
}
]
```
Fields:
- `entity1`, `entity2`: 1-based indices in `sequences`.
- `copy1`, `copy2`: optional 1-based copy indices.
- `position1`, `position2`: 1-based residue/ligand-part positions.
- `atom1`, `atom2`: atom names. Integer references are also accepted for mapped
SMILES or file ligands.
Use `entity1`/`entity2` for new inputs. The old `left_entity`/`right_entity`
style is accepted for compatibility.
## Unsupported `constraint`
The inference-only build ignores legacy `constraint` fields. Use
`covalent_bonds` for supported covalent links.
## Output layout
`opendde pred` writes:
```text
<out_dir>/<job_name>/seed_<seed>/predictions/
├── <job_name>_sample_<rank>.cif
├── <job_name>_summary_confidence_sample_<rank>.json
└── <job_name>_full_data_sample_<rank>.json # only when --need_atom_confidence true
```
The summary JSON includes confidence metrics such as `plddt`, `gpde`, `ptm`,
`iptm`, clash flags, and `ranking_score` when available.
|