"""
Rowan v2 API: MSA (Multiple Sequence Alignment) Workflow
Generate multiple sequence alignments for protein sequences.
"""
from typing import Annotated
import rowan
import json
def submit_msa_workflow(
initial_protein_sequences: Annotated[str, "JSON string list of protein sequences (e.g., '[\"MKLLV...\", \"MAHQR...\"]')"],
output_formats: Annotated[str, "JSON string list of output formats - must be 'colabfold', 'chai', or 'boltz' (e.g., '[\"colabfold\", \"chai\"]')"],
name: Annotated[str, "Workflow name for identification and tracking"] = "MSA Workflow",
folder_uuid: Annotated[str, "UUID of folder to organize this workflow. Empty string uses default folder"] = "",
max_credits: Annotated[int, "Maximum credits to spend on this calculation. 0 for no limit"] = 0
):
"""Submit a multiple sequence alignment (MSA) workflow using Rowan v2 API.
Args:
initial_protein_sequences: JSON string list of protein sequences (amino acid strings)
output_formats: JSON string list of desired output formats. Valid options: 'colabfold', 'chai', 'boltz'
name: Workflow name for identification and tracking
folder_uuid: UUID of folder to organize this workflow. Empty string uses default folder.
max_credits: Maximum credits to spend on this calculation. 0 for no limit.
Generates multiple sequence alignments for protein sequences using advanced
alignment algorithms optimized for structure prediction tools. Useful for:
- Protein structure prediction with AlphaFold2/ColabFold
- Structure prediction with Chai-1
- Structure prediction with Boltz-1
- Evolutionary analysis and homology modeling
Valid output formats:
- 'colabfold': MSA format for ColabFold/AlphaFold2 structure prediction
- 'chai': MSA format optimized for Chai-1 structure prediction
- 'boltz': MSA format optimized for Boltz-1 structure prediction
Returns:
Workflow object representing the submitted workflow
Examples:
# MSA for ColabFold structure prediction
result = submit_msa_workflow(
initial_protein_sequences='["MKLLVLGLLLAAAVPGTRAAQMSFKLIGTEYFTLQIRGRERFEMFRELN"]',
output_formats='["colabfold"]',
name="Insulin MSA for ColabFold"
)
# MSA for multiple prediction tools
result = submit_msa_workflow(
initial_protein_sequences='["MKTAYIAKQRQISFVKSHFSRQ"]',
output_formats='["colabfold", "chai", "boltz"]',
name="Multi-tool MSA"
)
# MSA for Chai-1 structure prediction
result = submit_msa_workflow(
initial_protein_sequences='["GSTLGRIADRDLLELDTLAAKVPSDGAKDLVTDIVNRQIYDG"]',
output_formats='["chai"]',
name="Chai-1 MSA"
)
This workflow can take 10-30 minutes depending on sequence length.
"""
# Parse initial_protein_sequences (always a string in MCP)
try:
parsed_sequences = json.loads(initial_protein_sequences)
except (json.JSONDecodeError, ValueError):
# Try comma-separated
if ',' in initial_protein_sequences:
parsed_sequences = [s.strip() for s in initial_protein_sequences.split(',') if s.strip()]
else:
parsed_sequences = [initial_protein_sequences.strip()]
# Parse output_formats
try:
parsed_formats = json.loads(output_formats)
except (json.JSONDecodeError, ValueError):
# Try comma-separated
if ',' in output_formats:
parsed_formats = [f.strip() for f in output_formats.split(',') if f.strip()]
else:
parsed_formats = [output_formats.strip()]
result = rowan.submit_msa_workflow(
initial_protein_sequences=parsed_sequences,
output_formats=parsed_formats,
name=name,
folder_uuid=folder_uuid if folder_uuid else None,
max_credits=max_credits if max_credits > 0 else None
)
# Make workflow publicly viewable
result.update(public=True)
return result
