similar-protein-retrieval

Similar Protein Retrieval

Retrieve proteins with similar structures, sequences, or from the same family using FoldSeek (structure) or MSA (sequence).

When to Use

• User provides a protein and wants to find similar proteins
• User asks for homologs or orthologs of a protein
• User wants proteins with similar 3D structure
• User wants to search by sequence similarity
• User provides UniProt ID, PDB ID, FASTA, or PDB file as input

Workflow

Step 1: Parse Input and Load Protein

Detect input type and load the protein appropriately.

import os
import requests
from open_biomed.data import Protein
from open_biomed.tools.tool_registry import TOOLS

def parse_input(user_input):
    """Parse input and return Protein object with structure info."""
    # Check if it's a file path
    if os.path.isfile(user_input):
        if user_input.endswith('.pdb'):
            return Protein.from_pdb_file(user_input), True, "pdb_file"
        elif user_input.endswith(('.fasta', '.fa')):
            with open(user_input) as f:
                seq = ''.join(l.strip() for l in f if not l.startswith('>'))
            return Protein.from_fasta(seq), False, "fasta_file"

    # Check if it's a UniProt ID (e.g., P0DTC2)
    if len(user_input) in [6, 10] and user_input[0].isalpha():
        return query_uniprot(user_input)

    # Check if it's a PDB ID (4 characters, e.g., 6LZG)
    if len(user_input) == 4 and user_input[0].isdigit():
        return query_pdb(user_input)

    # Assume it's a FASTA sequence
    return Protein.from_fasta(user_input), False, "fasta_string"

Step 2a: Query UniProt (if UniProt ID)

def query_uniprot(uniprot_id):
    """Query UniProt for sequence and PDB cross-references."""
    url = f"https://rest.uniprot.org/uniprotkb/{uniprot_id}?format=json"
    response = requests.get(url)
    data = response.json()

    sequence = data['sequence']['value']
    protein = Protein.from_fasta(sequence)
    protein.name = uniprot_id

    # Get PDB cross-references
    xrefs = data.get('uniProtKBCrossReferences', [])
    pdb_refs = [x['id'] for x in xrefs if x['database'] == 'PDB']

    has_structure = len(pdb_refs) > 0
    return protein, has_structure, "uniprot", {"pdb_refs": pdb_refs}