| """Format sequences for submission to the three online predictors. |
| |
| This is the 'link out' half of the hybrid model: CasMD writes |
| submission-ready files and prints the server URLs; the user pastes |
| into the web UIs manually. |
| """ |
| from __future__ import annotations |
| from dataclasses import dataclass |
| from pathlib import Path |
| import json |
|
|
|
|
| SERVER_URLS = { |
| "boltz": "https://tamarind.bio/boltz-2", |
| "protenix": "https://www.bioos.com/protenix", |
| "af3": "https://alphafoldserver.com", |
| } |
|
|
|
|
| @dataclass(frozen=True) |
| class PrepResult: |
| boltz_input: Path |
| protenix_input: Path |
| af3_input: Path |
|
|
|
|
| def prep_inputs( |
| *, |
| protein_fasta: str, |
| rna_sequence: str | None, |
| dna_sequence: str | None, |
| target_complement: str | None, |
| output_dir: Path, |
| job_name: str, |
| ) -> PrepResult: |
| """Write three submission files (one per backend) ready to paste.""" |
| output_dir = Path(output_dir) |
| output_dir.mkdir(parents=True, exist_ok=True) |
|
|
| protein_seq = _fasta_to_seq(protein_fasta) |
|
|
| |
| boltz_lines = ["sequences:"] |
| boltz_lines.append(" - protein:") |
| boltz_lines.append(" id: A") |
| boltz_lines.append(f" sequence: {protein_seq}") |
| chain_id = ord("B") |
| if rna_sequence: |
| boltz_lines.append(" - rna:") |
| boltz_lines.append(f" id: {chr(chain_id)}") |
| boltz_lines.append(f" sequence: {rna_sequence}") |
| chain_id += 1 |
| if dna_sequence: |
| boltz_lines.append(" - dna:") |
| boltz_lines.append(f" id: {chr(chain_id)}") |
| boltz_lines.append(f" sequence: {dna_sequence}") |
| chain_id += 1 |
| if target_complement: |
| boltz_lines.append(" - dna:") |
| boltz_lines.append(f" id: {chr(chain_id)}") |
| boltz_lines.append(f" sequence: {target_complement}") |
| boltz_path = output_dir / "boltz_input.yaml" |
| boltz_path.write_text("\n".join(boltz_lines) + "\n") |
|
|
| |
| protenix_seqs: list[dict] = [ |
| {"proteinChain": {"sequence": protein_seq, "count": 1}} |
| ] |
| if rna_sequence: |
| protenix_seqs.append({"rnaSequence": {"sequence": rna_sequence, "count": 1}}) |
| if dna_sequence: |
| protenix_seqs.append({"dnaSequence": {"sequence": dna_sequence, "count": 1}}) |
| if target_complement: |
| protenix_seqs.append({"dnaSequence": {"sequence": target_complement, "count": 1}}) |
| protenix_input = [{"name": job_name, "covalent_bonds": [], "sequences": protenix_seqs}] |
| protenix_path = output_dir / "protenix_input.json" |
| protenix_path.write_text(json.dumps(protenix_input, indent=2)) |
|
|
| |
| af3_seqs: list[dict] = [ |
| {"proteinChain": {"sequence": protein_seq, "count": 1, "useStructureTemplate": False}} |
| ] |
| if rna_sequence: |
| af3_seqs.append({"rnaSequence": {"sequence": rna_sequence, "count": 1}}) |
| if dna_sequence: |
| af3_seqs.append({"dnaSequence": {"sequence": dna_sequence, "count": 1}}) |
| if target_complement: |
| af3_seqs.append({"dnaSequence": {"sequence": target_complement, "count": 1}}) |
| af3_input = [ |
| { |
| "name": job_name, |
| "modelSeeds": [], |
| "sequences": af3_seqs, |
| "dialect": "alphafoldserver", |
| "version": 3, |
| } |
| ] |
| af3_path = output_dir / "af3_input.json" |
| af3_path.write_text(json.dumps(af3_input, indent=2)) |
|
|
| return PrepResult( |
| boltz_input=boltz_path, |
| protenix_input=protenix_path, |
| af3_input=af3_path, |
| ) |
|
|
|
|
| def _fasta_to_seq(fasta: str) -> str: |
| """Strip header lines from a FASTA, return concatenated sequence.""" |
| lines = [ln.strip() for ln in fasta.strip().splitlines()] |
| return "".join(ln for ln in lines if not ln.startswith(">")) |
|
|
