Team
A leading management team, backed by proven scientific founders.
Executive team
Scientific co-founders
We have a rich toolkit for editing the genome and transcriptome, but an equivalent framework for the proteome has remained elusive. UbiquiTx has built an AI-native platform for programmable modulation of the proteome — turning any protein target into one we can precisely modulate, not just knock out.
A single, modular construct: an AI-designed guide peptide, joined by a flexible linker to a swappable effector enzyme.
No binding pocket required. Guide peptides engage flat, non-conformational surfaces that PROTACs and molecular glues can't reach.
Swap the effector to degrade, stabilize, or relocalize the same target — not degradation alone.
Binding and function happen in the same step, so every hit from a screen is a functional one — nothing to validate twice.
Multiplexed peptide screening finds the best binders first. Combinatorial CLIP assembly then turns those winners into a functional cellular outcome — degraded, stabilized, or relocalized.
Multiple protein language models generate and rank candidates — not hand-engineered guesses.
Every cycle produces both binding and functional readouts — proprietary data no public database has.
Design–screen–measure–learn compounds — every round sharpens the next, across every program.
Protein models learn what binds. Almost nothing records what happened next. Because our screens read out function directly, every campaign generates exactly that missing data — at scale, across every program.
Delivers in-cell, mechanism-matched modulation of a nominated target, with dose-response, kinetics, and PTM/isoform selectivity — before partners commit capital.
Delivers a mechanism-matched go/no-go on a nominated target by addressing a protein directly.
Sequence-level and permanent. Powerful for somatic targets, but acts on the gene.
Knockdown only. Transient, with limited tissue reach.
Genetic tools answer a different question than the one a drug will face.
A pipeline in oncology and regenerative disease, advancing on non-dilutive funding alongside the validation business rather than competing with it for capital.
Built on a decade of peer-reviewed science from our founders.
Guide peptides designed with the SaLT&PepPr protein language model turn ubiquitin ligases into programmable degraders that selectively eliminate abnormally accumulated (cytosolic/nuclear) β-catenin in colorectal cancer cells — while leaving normal, membrane-bound β-catenin untouched. Validated in vitro and in vivo.
Read the paper →Peptide guides designed by protein language models (SaLT&PepPr, PepMLM, PepPrCLIP) are fused to the deubiquitinase OTUB1 to build "deubiquibodies" (duAbs) that rescue disease-relevant proteins — including p53, FOXP3, WEE1, and the disordered PAX3::FOXO1 fusion oncoprotein — by stripping their degradation-tagging ubiquitin. mRNA-lipid-nanoparticle delivery of a p53-targeting duAb restored p53 activity and triggered apoptosis in cancer cells.
Read the paper →PepPrCLIP pairs generative peptide design with a CLIP-style discriminator trained on peptide–protein pairs to design binders for targets with no fixed structure, including intrinsically disordered oncoproteins. Peptides generated this way, from sequence alone, inhibited an enzyme target and drove degradation of both β-catenin and the disordered SS18–SSX1 fusion protein found in synovial sarcoma.
Read the paper →SaLT&PepPr fine-tunes the ESM-2 protein language model to predict protein-protein interaction interfaces directly from sequence, then extracts guide peptides from those interfaces and fuses them to the CHIP E3 ligase domain to build programmable degraders — no structural information required. Degraders built this way knocked down β-catenin, 4E-BP2, and TRIM8 with high specificity, and TRIM8 degradation induced apoptosis in Ewing sarcoma cells.
Read the paper →A bacterial E3 ligase mimic (IpaH9.8, from Shigella flexneri) fused to synthetic binding domains degraded GFP and 15 other fluorescent-protein-tagged targets across a wide range of sizes and cellular compartments, then was redirected to disease-relevant targets SHP2, KRas, and ERK2 — delivered as a cell-penetrating peptide fusion or as mRNA, with efficient target depletion in cultured cells and in transgenic mice.
Read the paper →The foundational ubiquibody design: antibody fragments or fibronectin-based binding domains fused to a truncated CHIP E3 ligase redirect the ubiquitin-proteasome system to degrade a chosen target protein, with no need to know its natural interaction partners. Demonstrated by selectively depleting β-galactosidase and maltose-binding protein in mammalian cells while leaving natural CHIP substrates untouched — the "guide + ligase" architecture the platform builds on.
Read the paper →For partnership, licensing, and investment inquiries, we'd be glad to talk.
accounts@ubiquitx.com
