$npx -y skills add adaptyvbio/protein-design-skills --skill setupFirst-time setup for protein design tools. Use this skill when: (1) User is new and hasn't run any tools yet, (2) Commands fail with "file not found" or "modal: command not found", (3) Modal authentication errors occur, (4) User asks how to get started or set up the environment,
| 1 | # Setup Guide |
| 2 | |
| 3 | Help users get their environment ready to run protein design tools. |
| 4 | |
| 5 | ## Quick checklist |
| 6 | |
| 7 | Run through this checklist when a user encounters setup issues: |
| 8 | |
| 9 | | Step | Check | Fix | |
| 10 | |------|-------|-----| |
| 11 | | 1. Modal CLI | `modal --version` | `pip install modal` | |
| 12 | | 2. Modal auth | `modal token show` | `modal setup` | |
| 13 | | 3. biomodals | `ls biomodals/modal_*.py` | `git clone https://github.com/hgbrian/biomodals` | |
| 14 | | 4. Test | `cd biomodals && modal run modal_boltzgen.py --help` | See troubleshooting | |
| 15 | |
| 16 | ## Diagnosing issues |
| 17 | |
| 18 | ### Error: "modal: command not found" |
| 19 | |
| 20 | **Cause**: Modal CLI not installed. |
| 21 | |
| 22 | **Fix**: |
| 23 | ```bash |
| 24 | pip install modal |
| 25 | ``` |
| 26 | |
| 27 | Then restart the terminal or run `hash -r`. |
| 28 | |
| 29 | ### Error: "Permission denied" or "Unauthorized" |
| 30 | |
| 31 | **Cause**: Modal not authenticated. |
| 32 | |
| 33 | **Fix**: |
| 34 | ```bash |
| 35 | modal setup |
| 36 | ``` |
| 37 | |
| 38 | This opens a browser. Click "Authorize" to complete authentication. |
| 39 | |
| 40 | ### Error: "No such file or directory: modal_boltzgen.py" |
| 41 | |
| 42 | **Cause**: biomodals repository not cloned or not in correct directory. |
| 43 | |
| 44 | **Fix**: |
| 45 | ```bash |
| 46 | git clone https://github.com/hgbrian/biomodals |
| 47 | cd biomodals |
| 48 | ``` |
| 49 | |
| 50 | ### Error: "uvx: command not found" |
| 51 | |
| 52 | **Cause**: `uvx` is an optional wrapper from the `uv` package. It's not required. |
| 53 | |
| 54 | **Fix**: Run modal directly (recommended): |
| 55 | ```bash |
| 56 | modal run modal_boltzgen.py --help |
| 57 | ``` |
| 58 | |
| 59 | Or install uv if you prefer using uvx: |
| 60 | ```bash |
| 61 | pip install uv |
| 62 | ``` |
| 63 | |
| 64 | ## Full setup steps |
| 65 | |
| 66 | ### Step 1: Install Modal CLI |
| 67 | |
| 68 | ```bash |
| 69 | pip install modal |
| 70 | ``` |
| 71 | |
| 72 | Verify: `modal --version` |
| 73 | |
| 74 | ### Step 2: Authenticate Modal |
| 75 | |
| 76 | ```bash |
| 77 | modal setup |
| 78 | ``` |
| 79 | |
| 80 | This opens a browser. Click "Authorize". |
| 81 | |
| 82 | Verify: `modal token show` |
| 83 | |
| 84 | ### Step 3: Clone biomodals |
| 85 | |
| 86 | ```bash |
| 87 | git clone https://github.com/hgbrian/biomodals |
| 88 | cd biomodals |
| 89 | ``` |
| 90 | |
| 91 | Verify: `ls modal_*.py` should show files like `modal_boltzgen.py` |
| 92 | |
| 93 | ### Step 4: Test the Setup |
| 94 | |
| 95 | ```bash |
| 96 | cd biomodals |
| 97 | modal run modal_boltzgen.py --help |
| 98 | ``` |
| 99 | |
| 100 | Expected: Usage instructions appear showing `--input-yaml`, `--protocol`, `--num-designs` options. |
| 101 | |
| 102 | ## Common workflows after setup |
| 103 | |
| 104 | Once setup is complete, users can: |
| 105 | |
| 106 | ```bash |
| 107 | cd biomodals |
| 108 | |
| 109 | # Design binders with BoltzGen (requires YAML config) |
| 110 | modal run modal_boltzgen.py --input-yaml binder.yaml --protocol protein-anything --num-designs 50 |
| 111 | |
| 112 | # Generate backbones with RFdiffusion (official repo, not biomodals) |
| 113 | python run_inference.py inference.input_pdb=target.pdb contigmap.contigs=[A1-150/0 70-100] inference.num_designs=100 |
| 114 | |
| 115 | # Validate with Chai |
| 116 | modal run modal_chai1.py --input-faa designs.fasta |
| 117 | ``` |
| 118 | |
| 119 | ## GPU selection |
| 120 | |
| 121 | Set GPU with environment variable: |
| 122 | |
| 123 | ```bash |
| 124 | GPU=L40S modal run modal_boltzgen.py --input-yaml config.yaml --num-designs 50 |
| 125 | GPU=A100 modal run modal_chai1.py --input-faa complex.fasta |
| 126 | ``` |
| 127 | |
| 128 | | GPU | VRAM | Best For | |
| 129 | |-----|------|----------| |
| 130 | | T4 | 16GB | ProteinMPNN, ESM | |
| 131 | | A10G | 24GB | RFdiffusion, Chai | |
| 132 | | L40S | 48GB | BoltzGen, BindCraft | |
| 133 | | A100 | 40-80GB | Large complexes | |
| 134 | |
| 135 | ## Modal free tier |
| 136 | |
| 137 | Modal offers $30/month in free credits - enough for: |
| 138 | - ~500 BoltzGen designs |
| 139 | - ~2000 RFdiffusion backbones |
| 140 | - ~1000 Chai predictions |
| 141 | |
| 142 | --- |
| 143 | |
| 144 | **Full documentation**: See [Getting started](../../docs/getting-started.md) |