Day 10: Conserved Protein Domain Architecture: CDD, Pfam & InterProScan

1. The Modular Architecture of Proteins

In structural and evolutionary biology, a protein domain is an autonomous, conserved structural and functional unit that folds independently of the rest of the polypeptide chain, exhibits an independent evolutionary trajectory, and frequently mediates a distinct biochemical function (e.g. ATP binding, phosphotyrosine recognition, DNA binding, or transmembrane anchoring).

Through evolutionary processes such as exon shuffling, gene duplication, and retrotransposition, Nature combines elementary domains into diverse multidomain proteins. Consequently, classifying a protein by its domain architecture reveals its molecular function far more reliably than raw sequence identity.

2. Profile Hidden Markov Models (HMMs) in Domain Detection

Traditional consensus sequences fail to capture positional variability across diverse domain homologues. Profile HMMs (developed by Sean Eddy in HMMER) model an alignment column-by-column as a statistical Markov chain comprising three fundamental states:

• Match States ($M_k$): Probability distribution over the 20 amino acids at conserved consensus positions.
• Insert States ($I_k$): Probability of inserting one or more non-conserved amino acids between columns.
• Delete States ($D_k$): Probability of omitting a consensus residue (deletions).

$$\log \left( \frac{P(\text{Sequence} \mid \text{HMM})}{P(\text{Sequence} \mid \text{Null Model})} \right) \implies \text{Bit Score}$$

3. Production Python Script: Programmatic InterProScan REST API Query

#!/usr/bin/env python3
"""
Day 10 Practical: Automated Domain Annotation via InterProScan API
Submits a protein sequence and extracts Pfam, SMART, and CDD domain matches.
"""

import urllib.request
import urllib.parse
import json
import time
import sys

INTERPRO_API = "https://www.ebi.ac.uk/Tools/services/rest/iprscan5"

def submit_interpro_job(sequence_fasta: str, email: str):
    """Submit FASTA sequence to EBI InterProScan REST service."""
    params = urllib.parse.urlencode({
        "email": email,
        "title": "OmicsDomainScan",
        "goterms": "true",
        "pathways": "true",
        "sequence": sequence_fasta
    }).encode("utf-8")
    
    req = urllib.request.Request(f"{INTERPRO_API}/run", data=params)
    with urllib.request.urlopen(req) as resp:
        job_id = resp.read().decode("utf-8").strip()
    print(f"InterProScan Job Submitted. Job ID: {job_id}")
    return job_id

def poll_and_fetch_results(job_id: str):
    """Poll job status until FINISHED, then download JSON output."""
    status_url = f"{INTERPRO_API}/status/{job_id}"
    while True:
        with urllib.request.urlopen(status_url) as resp:
            status = resp.read().decode("utf-8").strip()
        print(f"Job Status: {status}...")
        if status == "FINISHED":
            break
        elif status in ["ERROR", "FAILURE", "NOT_FOUND"]:
            print("Job failed on EBI server.", file=sys.stderr)
            return None
        time.sleep(5)
        
    result_url = f"{INTERPRO_API}/result/{job_id}/json"
    with urllib.request.urlopen(result_url) as resp:
        return json.loads(resp.read().decode("utf-8"))

if __name__ == "__main__":
    # Human p53 DNA-binding core domain sequence
    p53_fragment = (
        ">P53_CORE_DOMAIN\n"
        "MPEAAPPVAPAPAAPTPAAPAPAPSWPLSSSVPSQKTYQGSYGFRLGFLHSGTAKSVTCTYSPALNKMFCQLAKTCP"
        "VQLWVDSTPPPGTRVRAMAIYKQSQHMTEVVRRCPHHERCSDSDGLAPPQHLIRVEGNLRVEYLDDRNTFRHSVVVP"
        "YEPPEVGSDCTTIHYNYMCNSSCMGGMNRRPILTIITLEDSSGNLLGRNSFEVRVCACPGRDRRTEEENLRKKGEPH"
        "HELP"
    )
    print("Submitting protein sequence to InterProScan API...")
    # Example simulated output inspection
    print("Identified Domain: P53 DNA-binding domain (Pfam: PF00870, InterPro: IPR011615)")
    print("Coordinates: Residues 102 - 292 | E-value: 2.4e-68")