File size: 3,848 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
# OpenDDE Tutorial

A short walkthrough using files in [`examples/`](../examples). For install and
runtime data setup, see [inference_instructions.md](./inference_instructions.md)
or [docker_installation.md](./docker_installation.md).

## 1. Check the environment

Run commands from the repository root:

```bash
opendde doctor
export OPENDDE_ROOT_DIR=/path/to/opendde_data
```

Prediction needs:

```text
$OPENDDE_ROOT_DIR/checkpoint/opendde.pt
$OPENDDE_ROOT_DIR/common/
```

The released general-purpose checkpoint is
[`opendde.pt`](https://huggingface.co/aurekaresearch/OpenDDE/resolve/main/opendde.pt).
For antibody-antigen (ABAG) complexes, use the ABAG-optimized
[`opendde_abag.pt`](https://huggingface.co/aurekaresearch/OpenDDE/resolve/main/opendde_abag.pt).
Place them under `$OPENDDE_ROOT_DIR/checkpoint/`, preserving the filenames. Pass
`opendde_abag.pt` directly with `--load_checkpoint_path` for ABAG runs.

```bash
mkdir -p "$OPENDDE_ROOT_DIR/checkpoint"
curl -L \
  -o "$OPENDDE_ROOT_DIR/checkpoint/opendde.pt" \
  https://huggingface.co/aurekaresearch/OpenDDE/resolve/main/opendde.pt
```

Template/RNA-MSA preprocessing also needs `hmmer`; template inference may need
`kalign`.

## 2. Compatibility prediction

This disables external features and keeps the standard step/cycle counts.
Inference defaults to `fp32` and `auto` triangle kernels (PyTorch on CPU), so no
extra dtype or kernel flags are needed:

```bash
opendde pred \
  -i examples/input.json \
  -o ./output \
  -n opendde_v1 \
  --use_msa false \
  --use_template false \
  --use_rna_msa false \
  --sample 1 \
  --step 200 \
  --cycle 10
```

Outputs go to:

```text
output/<job_name>/seed_<seed>/predictions/
```

## 3. Input JSON basics

OpenDDE input is a list of jobs:

```json
[
  {
    "name": "tiny",
    "sequences": [
      {
        "proteinChain": {
          "sequence": "ACDEFGHIK",
          "count": 1
        }
      }
    ]
  }
]
```

`covalent_bonds` is optional here and can be left out; it is only needed to
declare explicit covalent links between entities.

Entity keys include `proteinChain`, `dnaSequence`, `rnaSequence`, `ligand`, and
`ion`. Full schema: [infer_json_format.md](./infer_json_format.md).

Convert a PDB/CIF instead of writing JSON by hand:

```bash
opendde json -i examples/7pzb.pdb -o ./output --altloc first
```

## 4. Use precomputed MSA/template features

[`examples/examples_with_template/example_9fm7.json`](../examples/examples_with_template/example_9fm7.json)
already contains `pairedMsaPath`, `unpairedMsaPath`, and `templatesPath`:

```bash
opendde pred \
  -i examples/examples_with_template/example_9fm7.json \
  -o ./output \
  -n opendde_v1 \
  --use_msa true \
  --use_template true \
  --use_rna_msa false
```

## 5. Generate MSA/template features

For an input without MSA/template paths:

```bash
opendde prep -i examples/example_without_msa.json -o ./output
```

This writes an updated JSON next to the input. Predict from that updated JSON:

```bash
opendde pred \
  -i examples/example_without_msa-final-updated.json \
  -o ./output \
  -n opendde_v1 \
  --use_msa true \
  --use_template true \
  --use_rna_msa false
```

For protein MSA only, use `opendde msa`. For protein MSA + template only, use
`opendde mt`.

## 6. RNA MSA example

[`examples/examples_with_rna_msa/example_9gmw_2.json`](../examples/examples_with_rna_msa/example_9gmw_2.json)
contains a precomputed RNA MSA:

```bash
opendde pred \
  -i examples/examples_with_rna_msa/example_9gmw_2.json \
  -o ./output \
  -n opendde_v1 \
  --use_rna_msa true
```

To generate RNA MSA for your own RNA input, run `opendde prep` first.

## More details

- [Inference instructions](./inference_instructions.md)
- [Input JSON format](./infer_json_format.md)
- [MSA/template/RNA-MSA pipeline](./msa_template_pipeline.md)
- [Kernel options](./kernels.md)