CORE TECHNOLOGY 核心技术平台

The DTBL closed-loop — an AI-powered full-cycle R&D engine DTBL 闭环 —— AI 驱动的全链路新药研发引擎

Design, Test, Build, Learn. Every stage is powered by proprietary AI models and validated by atomic-level Cryo-EM structural biology — closing the loop between computation and wet-lab reality. Design(设计)、Test(验证)、Build(合成)、Learn(学习)。每个环节均由自研 AI 模型驱动, 并通过冷冻电镜原子级解析进行实验验证 —— 真正打通干实验与湿实验的闭环。

The DTBL Framework DTBL 框架

A continuously learning drug discovery loop 持续学习的新药发现闭环

D
Design 设计
AI molecular design — VAE / Diffusion generative models with ADMET evaluation. AI 分子设计 —— VAE / 扩散生成模型 + ADMET 成药性评估。
T
Test 验证
Cryo-EM (< 2.5 Å), dual-cell HTS, and in-vivo models. 冷冻电镜(< 2.5 Å)、双细胞高通量筛选与体内模型。
B
Build 合成
Automated synthesis — AI-driven retrosynthesis with robotics. 自动化合成 —— AI 驱动的逆合成分析与机器人实验。
L
Learn 学习
Data feedback loop — dry-wet data pool and active model iteration. 数据反馈学习 —— 干湿实验数据池驱动模型主动迭代。
AI-Driven Drug Discovery (AIDD) AI 驱动药物发现(AIDD)

Published, benchmarked, production-ready algorithms 可复用的 AI 算法矩阵

Our AIDD stack spans target discovery, molecular generation, and druggability scoring — each algorithm peer-reviewed and validated against experimental endpoints. 我们的 AIDD 技术栈涵盖靶点发现、分子生成与成药性评估 —— 每个算法均已在同行评审期刊发表,并通过实验端点验证。

🕸️

NBI / SDTNBI / wSDTNBI

Network-Based Inference algorithms for target discovery. The foundational paper has 770+ citations (PLoS Comput. Biol. 2012). 基于网络的靶点推断算法族。基础论文被引 770+ 次 (PLoS Comput. Biol. 2012)。

🧩

ComplexDnet

Complex-disease target prediction engine, validated in NASH (RORγt target, 10nM lead, published in J. Med. Chem. 2025). 复杂疾病靶标预测引擎,已在 NASH(RORγt 靶点,10nM 先导物)中验证 —— 论文发表于 J. Med. Chem. 2025。

🔷

SGEDiff

SE(3)-equivariant GNN + diffusion model for 3D molecular generation, validated on CDK2, ABL1, AKT1, PPARα. SE(3) 等变 GNN + 扩散模型的三维分子生成方法, 已在 CDK2、ABL1、AKT1、PPARα 等靶点验证。

📊

KG-CNNDTI

Knowledge-graph-enhanced drug-target interaction prediction. Screened 13 Alzheimer's-related targets to discover 40 natural product hits. 知识图谱增强的药物-靶点相互作用预测。已筛选 13 个阿尔茨海默病关键靶标, 发现 40 个天然产物命中。

💊

admetSAR 3.0

Open ADMET prediction platform with 2M+ uses worldwide. Benchmark-grade coverage across absorption, distribution, metabolism, excretion, and toxicity. 开放 ADMET 预测平台,全球使用量突破 200 万次。 覆盖吸收、分布、代谢、排泄与毒性全端点预测。

🎯

CLaSP

Druggability scoring system that ranks candidate molecules by clinical success potential. 成药性打分系统,依据临床成功潜力对候选分子进行排序。

TransProtein Peptide Platform TransProtein 多肽设计平台

Full-cycle AI peptide engineering AI 全周期多肽设计体系

Six modules covering linear peptides, cyclic peptides, small-molecule-to-peptide conversion, BBB-penetrating peptides, multi-target peptides, and peptide conjugates. 六大模块覆盖线性多肽、环肽、小分子转肽、跨血脑屏障多肽、多靶点多肽与多肽偶联。

TransCyclepeptide

Cyclic peptide design with enhanced stability and bioavailability. 环肽设计,提升稳定性与生物利用度。

TransBBBpeptide

BBB-penetrating peptides for CNS disease targeting — 10× penetration improvement. 跨血脑屏障多肽,用于中枢神经系统疾病 —— 穿透率提升 10 倍。

Transmolecule + PharmaPepGen

Small-molecule-to-peptide conversion with AI-guided sequence optimization. 小分子转多肽转化,通过 AI 引导序列优化。

Produalnet

Multi-target peptide design for complex disease polytherapy. 多靶点多肽设计,面向复杂疾病的多机制联合治疗。

Experimental Validation 实验验证能力

Atomic-level structural biology — < 2.5 Å Cryo-EM resolution 原子级结构生物学 —— 冷冻电镜 < 2.5 Å 分辨率

Our integrated Cryo-EM pipeline captures binding modes at atomic resolution, feeding structural ground-truth back into our AI models and dramatically improving hit rate. 我们整合的冷冻电镜流水线在原子分辨率下捕获结合模式, 将结构真值反哺 AI 模型,显著提升命中率。

< 2.5 Å
Cryo-EM Resolution 冷冻电镜分辨率
10M+
Compound Library 分子库规模
10+
Core Algorithms 核心算法
2M+
admetSAR Platform Uses admetSAR 平台使用量
Competitive Advantage 竞争护城河

Three reinforcing layers no single competitor can replicate 三层叠加护城河,任何单一竞争者均无法复制

Pure-computation companies lack experimental validation depth. Traditional pharma lacks AI speed. GenTide's moat is the combination. 纯计算公司缺乏实验验证深度,传统药企缺乏 AI 速度。 GenTide 的护城河恰在两者的结合。

01

Speed Asymmetry — AI Platform 速度不对称 — AI 平台

DTBL compresses the target-to-PCC timeline from the industry average of 3–4 years down to 1.5–2 years, saving ~75% of per-project R&D cost. This isn't "doing the same work faster" — it's a fundamentally different iteration speed. DTBL 将靶点到 PCC 的周期从行业平均 3–4 年压缩至 1.5–2 年, 节省单个项目约 75% 的 R&D 成本。这不是"同样的工作做快一点", 而是从根本上改变了化合物优化的迭代速度。

40% ↓ timeline研发周期 70% ↓ cost成本 50% ↑ hit rate命中率
02

Precision Asymmetry — Cryo-EM 精度不对称 — 冷冻电镜

Our <2.5 Å Cryo-EM pipeline provides the atomic-level structural ground truth that feeds back into AI models — closing the loop that pure-computation companies cannot close. Most AI drug discovery companies' shortfall is precisely the absence of high-quality experimental validation. 我们 <2.5 Å 的冷冻电镜管线提供原子级结合结构真值,反哺 AI 模型—— 打通了纯计算公司无法打通的闭环。大多数 AI 制药公司的短板恰恰在于 缺乏高质量的实验验证能力,导致计算预测与真实活性之间存在巨大落差。

<2.5 Å resolution分辨率 20–50× targeting靶向富集
03

First-mover Asymmetry — Original Target 先发不对称 — 原创靶点

The β2-AR anti-aging target is a GenTide-exclusive asset: 3–5 year head start, a Nature Aging mechanistic publication, proprietary p16 reporter mice HTS system, and a complete in-vivo validation loop. Competitors cannot fast-follow with money alone. β2-AR 抗衰老靶点是 GenTide 独有资产:3–5 年先发窗口,Nature Aging 机制发表, 独家 p16 报告基因小鼠 HTS 体系,以及完整的体内验证闭环。 竞争对手无法仅靠砸钱快速跟进。

3–5 yr head start先发优势 $500B+ market潜在市场
IP Protection 知识产权保护

Triple patent moat — >20-year protection 三重专利护城河 —— 保护期超 20 年

🤖

AI Algorithm Patents AI 算法专利

Method patents on generative design models and binding-site conjugate design models. 生成式设计模型与结合位点偶联设计模型的方法专利。

🏗️

Scaffold Patents 结构骨架专利

Cryo-EM binding-pocket-derived Markush scaffolds, providing broad structural coverage. 通过冷冻电镜结合口袋确定的 Markush 骨架,提供宽泛的结构覆盖。

⚗️

CMC Conjugation Process CMC 偶联工艺专利

Novel linker design and site-specific conjugation process patents for AI-designed conjugates. 新型连接子设计与定点偶联工艺专利,覆盖 AI 设计偶联体。

🌍

PCT + Divisional Strategy PCT + 分案策略

Coverage in China, US, EU, and Japan via PCT filing with divisional applications. 20+ year protection per asset. 通过 PCT 分案策略覆盖中国 / 美国 / 欧盟 / 日本,每项资产保护期超 20 年。