MOSAIC™ Platform

MOSAIC™mRNA Display based Optimization and Selection with Artificial Intelligence for Cyclic Peptide

The MOSAIC™ Platform is built for the discovery and optimization of noncanonical amino acid-enabled cyclic peptides. It connects programmable genetic codes, ultra-large mRNA display selection, NGS sequence analysis, structural modeling, AI candidate design, synthesis, and experimental validation in one continuous discovery system. Computation and experiment are not parallel tracks: selection, counter-selection, and validation data enter the next model and molecular-design cycle to progressively improve hit quality, selectivity, and developability.

A Dual-Engine Cyclic Peptide Discovery System

MOSAIC™ Platform is Apollomics’ technology platform for the discovery and optimization of noncanonical amino acid-enabled cyclic peptides. It connects high-throughput mRNA display selection, programmable noncanonical amino acid chemistry, AI and structure-driven molecular design, and experimental validation so that experimental data can continuously inform the next design cycle.

MOSAIC cyclic peptide discovery workflow integrating computational design and mRNA display
AI Computational Design

Structure modeling · Candidate generation · Prioritization

Experimental data continuously informs design
mRNA Display Discovery

Ultra-large libraries · Iterative selection · Sequence analysis

Reprogramming the Genetic Code to Expand Cyclic Peptide Chemical Space

Selected codons are assigned to defined amino acid chemistry, translated into traceable mRNA–peptide fusions, and cyclized through programmable strategies to create diverse noncanonical cyclic peptide libraries.

Genetic code reprogramming workflow from aminoacyl-tRNA and mRNA-peptide fusion to a noncanonical cyclic peptide library
  1. 1

    Genetic Code Reprogramming

    Reassign selected codons to target noncanonical amino acids.

  2. 2

    Charged tRNA

    Load the target amino acid onto its corresponding tRNA enzymatically or chemically.

  3. 3

    In Vitro Translation & mRNA–Peptide Fusion

    Translate the reprogrammed message and establish a traceable genotype–phenotype linkage.

  4. 4

    Cyclized Library Construction

    Use head-to-tail, side-chain, or other programmable strategies to generate cyclic peptide libraries.

AI-Driven Peptide Developability and Oral-Potential Optimization

AI integrates cyclic-peptide conformation, potency, selectivity, solubility, proteolytic stability, polarity–lipophilicity balance, and permeability into multi-parameter candidate ranking. Experimental data continually refine the design direction to improve overall developability, while gastrointestinal stability and intestinal epithelial permeability provide evidence for oral-development potential.

AI combines IC50, microplate, solubility, stability, and permeability assays to optimize cyclic-peptide developability and assess oral potential