Autoimmune

TYK2

CHEMBL2148 (= JAK3) · real TYK2 = CHEMBL3553

VALIDATED

First production pipeline to combine Boltz-2 + ADMET-AI + ROBOGATE failure-boundary mapping on a single target.

데이터 무결성 / Data-lineage: This program was seeded from CHEMBL2148, which is in fact JAK3 (real TYK2 = CHEMBL3553). The R1 wet-lab hits are validated potent JAK-family / pan-JAK catalytic binders that cross-bind TYK2 — TYK2 selectivity is not yet proven. TYK2/JAK selectivity panel in progress.

ADMET Passed

6,478

Pareto Rank-1 Count

500

CRO

3 compounds — R1 3/3 active @ ~2 nM

Stage

Gate B in progress

Pareto Rank-1 is a front-membership count, not a quality ranking — the front is near-saturated, so it does not discriminate quality. Meaningful selection uses consensus + wet-lab.

Mechanism

ATP-competitive catalytic-site binder (R1). Distinct from deucravacitinib's allosteric JH2 pseudokinase mechanism — a novel mechanistic angle, pending selectivity confirmation.

Precedent

Deucravacitinib (BMS, approved 2022) · Zasocitinib (Takeda, Phase III)

Indications

Psoriasis · Rheumatoid arthritis · Crohn's disease

Milestones

2026-W09

Gate A assessment — 5 PASS · 1 MARGINAL · 2 FAIL → PIVOT

2026-W13

Top-3 compounds submitted to CRO (Eurofins Panlabs)

2026-W18

CRO IC50 readback expected (Gate B)

Dashboard →Browse compounds →Methodology →