Protein Function Prediction

Predict functional annotations and properties for proteins from their amino acid sequences using the BioT5 model.

When to Use

• You have a protein FASTA sequence and need to understand its biological role
• You want to identify enzyme function, pathway involvement, or molecular mechanisms
• You need quick functional insights without experimental data
• You're characterizing novel or unannotated protein sequences

Workflow

from open_biomed.data import Protein, Text
from open_biomed.core.pipeline import InferencePipeline

# Create protein from FASTA sequence
protein = Protein.from_fasta("YOUR_AMINO_ACID_SEQUENCE")

# Create the question for functional annotation
question = Text.from_str(
    "Inspect the protein sequence and offer a concise description of its properties."
)

# Load the BioT5 model for protein question answering
pipeline = InferencePipeline(
    task="protein_question_answering",
    model="biot5",
    model_ckpt="./checkpoints/server/protein_question_answering_biot5.ckpt",
    device="cuda:0"
)

# Run inference to get functional annotation
outputs = pipeline.run(protein=protein, text=question)
function_description = outputs[0][0].str
print(function_description)

See examples/basic_example.py for a complete runnable script.

Expected Outputs

The model returns a text description that typically includes:

Output Component	Example
Enzyme name	Phosphoribosylformylglycinamidine synthase
Biological pathway	Purine biosynthesis pathway
Catalytic activity	FGAR to FGAM conversion
Complex membership	Part of FGAM synthase complex (PurQ, PurL, PurS)
Mechanism details	ATP-dependent, glutamine amidotransferase activity

Example Output

Part of the phosphoribosylformylglycinamidine synthase complex involved in the purines biosynthetic pathway. Catalyzes the ATP-dependent conversion of formylglycinamide ribonucleotide (FGAR) and glutamine to yield formylglycinamidine ribonucleotide (FGAM) and glutamate.

Input Formats

The skill accepts protein sequences in FASTA format (amino acid string):

# From raw sequence string
protein = Protein.from_fasta("MRVGVIRFPGSNCDRDVHHVLELAGAEPEYVWW...")

# From UniProt (get sequence first)
from open_biomed.tools.tool_registry import TOOLS
tool = TOOLS["protein_uniprot_request"]
protein, _ = tool.run(accession="P00533")  # Example: EGFR

Error Handling

Error	Cause	Solution
FileNotFoundError	Model checkpoint not found	Download checkpoint to ./checkpoints/server/
CUDA out of memory	GPU memory insufficient	Use smaller batch or CPU device
Sequence too long	Exceeds 512 amino acid limit	Truncate sequence or use sliding window

Model Details

• Model: BioT5 (protein-text foundation model)
• Max sequence length: 512 amino acids
• Inference time: ~2-3 seconds per sequence on GPU
• Capabilities: Function prediction, property description, pathway annotation

Limitations

• Sequences longer than 512 residues are truncated
• Model trained on known proteins; novel folds may have lower accuracy
• Does not predict 3D structure or binding sites (use protein_folding or protein_binding_site_prediction tools)

Related Skills

• protein-structure-design-boltzgen: For 3D structure prediction
• protein-mutation-analysis: For mutation effect prediction
• uniprot-query: For retrieving protein metadata from UniProt