Protein Design for Antimicrobials

One-Sentence Definition

Computational protein design creates new amino-acid sequences/structures (often AI-assisted) intended as antimicrobials, binders, enzymes, or sensors that did not evolve in nature.

Simple Explanation

Instead of only finding antibiotics in soil microbes, we design proteins that stick to pathogen targets or break resistance enzymes.

Detailed Scientific Explanation

  • Leaders/tools lineage: Rosetta / RoseTTAFold / RFdiffusion-class methods — David Baker
  • Often combined with AlphaFold in Microbiology for structure prediction of designs
  • Modalities: antimicrobial peptides (AMPs), receptor traps, lysins, catalytic degraders of antibiotics? (research), diagnostic binders
  • Must optimize: potency, specificity, stability, manufacturability, toxicity, immunogenicity, resistance emergence

Mechanism

Specify target structure/function → generative model proposes backbones/sequences → filter in silico → synthesize → microbial assays → iterate.

Clinical Importance

  • Mostly preclinical; future niche for MDR pathogens and rapid-response antivirals
  • Regulatory path longer than small molecules in many cases

Research Importance

Diagnostic Relevance

  • Designed binder proteins → biosensors / capture reagents

AMR Relevance

  • Direct: novel agents against MDR organisms
  • Indirect: enzymes that restore antibiotic activity (e.g., β-lactamase inhibitors as proteins — conceptual)

Active Recall Questions

  1. Design vs prediction — difference?
  2. Why is off-target toxicity a big filter for AMPs?
  3. Which history figure anchors computational design in this vault?

Connections