100+全球先锋领袖现场开讲、50+顶尖新基建机构抢先入驻、1000+产业精英面对面链接,中国ADC和核药产业界规格最高、影响力最大、汇聚创新力量最全的年度品牌盛会!
英文版原文
(上下滑动查看更多)
ARTICLE | EMERGING COMPANY PROFILE
Epic ip hAI: Reading disease states from blood’s epigenetic fingerprints Chinese newco making a liquid biopsy thatdecodes chromatin marks to map tissues and outcomes
BY DANIELLE GOLOVIN, SENIOR BIOPHARMA ANALYST March 6, 2026 6:18 PM GMT+1
EpiciphAI is developing a liquid biopsy that reads histonemarks on cell-free chromatin to identify a fragment’s tissue of origin and distinguish disease-associated states within thattissue. Still at the pre-seed stage, the company sees an initial application in diagnosing and monitoring cancers in tissues that are difficult to biopsy.
Most epigenetic liquid biopsies read methylation marks written directly on DNA; EpiciphAI instead reads chemical tags on histone proteins that package the DNA, which may capture additional information about chromatin state and tissue pathology.When cells die, they shed cell-free chromatin fragments into the bloodstream. These fragments carry histone modifications — chemical tags on the histone proteins that DNA wraps around. Different tissues have unique patterns of histone modifications that act as molecular "fingerprints."
EphiciphAI’s technology, cf-EpiTracing, originated in the lab of Aibin He at Peking University. “As a developmental biologist, initially he was trying to develop a tool to trace the lineage of individual cells,” EphiciphAI CEO Jun Bao told BioCentury.
“Then he realized, hey, if I can trace individual cells, then I can know where the cell is coming from; I don't have to go to the individual organ to take the sample.”
He’s lab first took tissues from healthy volunteers or primary cells from cell banks and fragmented the cell-free DNA into 200 base pair fragments. They then mapped the histone modifications in the genome for three million chromatin fragments across 65 tissues to build a reference epigenome library.
Next, He’s group used proprietary algorithms to identify the set of histone modifications as the minimal panel that still recapitulated tissue and disease specificity, which turned out to be seven.
By focusing on seven histone modifications, each fragment can be represented as a seven-digit binary code, where each digit records whether a particular mark is present (1) or absent (0). That yields 27=128 possible patterns, but fragments carrying none of the seven marks (0000000) are never captured by any antibody, leaving 127 distinct
histone-mark combinations per fragment.
The epigenetic fingerprint depends not just on which marks are present, but also on where in the genome each fragment sits. Each of the ~3 million 200 bp fragments per sample is assigned a coordinate in the genome based on its DNA sequence, and the combination of genomic location plus the 127-mark pattern creates an astronomically large information space — enough to uniquely fingerprint every tissue and cell state in the body.
He's team used the 65-tissue epigenome reference datasets to identify recurring combinations of histone marks at specific genomic regions that characterize each tissue type and distinguish healthy from diseased states. They termed these tissue-specific patterns integrated chromatin states (ICS).
Finally, they built machine-learning models that compare a patient's blood-derived ICS fingerprints to this reference library, allowing the system to infer which tissues the
fragments came from and whether those tissues show signatures of specific diseases or stages.
His Peking University team published the details of cf- EpiTracing in Nature this week.
The paper highlighted several major capabilities: a) blood-based chromatin signatures enable earlier detection of lesions and can reduce reliance on invasive diagnostic procedures; b) different diseases affecting the same organ, such as inflammatory bowel disease and colorectal cancer, produce distinct ICS fingerprints that allow cf-EpiTracing to separate inflammatory from malignant processes non-invasively; c) age-related ICS patterns reveal that tissues at different ages show distinct epigenetic states, suggesting potential for monitoring organ aging and age-related disease risk, and) cf-EpiTracing can detect tissue involvement in secondary organs before it becomes clinically evident.
For example, liver tissue signature scores were elevated in many colorectal cancer patients even though only three had a clinical diagnosis of liver metastasis at the time of sampling. Similarly, in lymphoma patients, the platform flagged extranodal involvement in organs such as breast, pancreas and spleen that correlated with known patterns of disease spread.
“You'll be able to pick up micrometastasis, even before you can see it with your eye with pathology methods,” said Bao.
Bao said the company is planning to raise seed money and use it for “filing global patents, technology transfer agreements with research institutes, setting up company operations, and starting a few pilot studies.”
He envisions the technology will be first used to diagnose cancers and monitor disease progression for tissues that are difficult to biopsy. “We also think the technology can be used to screen therapeutics, for example, longevity related to tissue senescence, by collaborating with therapeutic companies.Lastly, using it as a more reliable measurement for minimum residual disease in cancer treatment, also by collaborating with cancer therapeutic companies,” said Bao.
In addition to the convenience of avoiding internal organ biopsies, Bao highlighted the cost effectiveness of the method. “The cost of this is really low,” he said. “It's about 500 RMB, less than $100.”
COMPANY PROFILE
Epic ip hAI
Chengdu and Beijing, China
Technology: Disease-agnostic liquid biopsy that relies on the fact that healthy and diseased cells carry distinct epigenetic "fingerprints"
Origin of technology: Peking University
Disease focus: Various (diagnostic is agnostic to disease)
Clinical status: Discovery
Founded: 2025 by Aibing He and Apuri BioVenture
Academic collaborators: Peking University
Corporate partners: None
Number of employees: 3
Funds raised: $4.2 million planned seed round
Investors: Undisclosed
CEO: Jun Bao
Issued Patents: None issued
来源:BIOCENTURY
中国新创企业研发液体活检技术,解码染色质标记以绘制组织图谱并预测疾病结局
EpiciphAI公司正研发一款液体活检产品,通过检测无细胞染色质上的组蛋白标记,确定染色质片段的组织来源,并区分该组织中与疾病相关的状态。目前该公司仍处于种子前阶段,其技术的首个应用场景为难以进行活检的组织的癌症诊断与病情监测。
大多数表观遗传液体活检技术检测的是直接标记在DNA上的甲基化修饰,而EpiciphAI则选择检测包裹DNA的组蛋白上的化学修饰标记,这类标记或能捕捉到更多关于染色质状态和组织病理特征的信息。
细胞凋亡时,会将无细胞染色质片段释放到血液中。这些片段带有组蛋白修饰——即DNA缠绕的组蛋白上的化学修饰标记。不同组织拥有独特的组蛋白修饰模式,这些模式可作为分子“指纹”。
EpiciphAI的cf-EpiTracing技术源自北京大学何爱彬教授的实验室。EpiciphAI首席执行官包骏在接受《生物世纪》采访时表示:“作为一名发育生物学家,何教授最初的研究目标是开发一种追踪单个细胞谱系的工具。后来他意识到,既然能追踪单个细胞,就能确定细胞的组织来源,无需再对各个器官进行侵入式取样。”
包骏称:“我们能检测到微转移灶的存在,甚至早于病理检测手段的可见阶段。”
何教授的团队首先获取健康志愿者的组织样本或细胞库的原代细胞,将无细胞DNA切割为200个碱基对的片段。随后,他们对65种组织中300万个染色质片段的全基因组组蛋白修饰情况进行分析,构建了表观基因组参考文库。
接下来,研究团队利用自主研发的算法,筛选出一组足以体现组织和疾病特异性的核心组蛋白修饰标记,最终确定了7种标记。
通过聚焦这7种组蛋白修饰,每个染色质片段可被编码为7位二进制代码,每位数字代表对应标记的存在(1)或缺失(0)。理论上可产生128种组合模式,但不携带任何7种标记的片段(0000000)无法被任何抗体捕获,因此每个片段实际可产生127种独特的组蛋白标记组合。
表观遗传指纹的特征不仅取决于组蛋白标记的种类,还与片段在基因组中的位置相关。研究人员会根据DNA序列,为每个样本中约300万个200碱基对的片段分配基因组坐标,基因组位置与127种标记模式的组合形成了海量的信息空间,足以对人体所有组织和细胞状态进行独特的指纹标记。
研究团队利用65种组织的表观基因组参考数据集,识别出特定基因组区域中反复出现的组蛋白标记组合,这些组合是各组织类型的特征标志,也能区分组织的健康与病变状态。研究人员将这些组织特异性模式命名为整合染色质状态(ICS)。
最终,团队构建了机器学习模型,将患者血液中的整合染色质状态指纹与参考文库进行比对,通过该系统可推断染色质片段的组织来源,以及这些组织是否呈现特定疾病或疾病特定阶段的特征。
何教授在北京大学的研究团队本周于《自然》杂志发表了cf-EpiTracing技术的详细研究成果。
该论文揭示了该技术的四大核心能力:第一,基于血液的染色质特征可实现病变的早期检测,减少对侵入式诊断手段的依赖;第二,同一器官受不同疾病影响时(如炎症性肠病和结直肠癌),会产生独特的整合染色质状态指纹,cf-EpiTracing技术可通过非侵入式手段区分炎症性病变与恶性病变;第三,与年龄相关的整合染色质状态模式表明,不同年龄段的组织呈现出独特的表观遗传状态,这意味着该技术有望用于监测器官衰老及年龄相关疾病的发病风险;第四,cf-EpiTracing技术能在继发性器官的组织受累表现出临床症状前,实现早期检测。
例如,许多结直肠癌患者的肝组织特征评分显著升高,而取样时仅有3名患者被临床诊断为肝转移。同样,在淋巴瘤患者中,该检测平台发现了乳腺、胰腺、脾脏等器官的结外受累迹象,且与已知的疾病扩散模式高度吻合。
包骏表示,公司计划启动种子轮融资,所筹资金将用于“申请全球专利、与研究机构签订技术转让协议、搭建公司运营体系以及开展多项试点研究”。
他设想该技术的首个应用场景为难以活检组织的癌症诊断和疾病进展监测。包骏称:“我们认为,通过与制药企业合作,该技术还可用于药物筛选,例如研发与组织衰老相关的长寿药物。此外,也可与肿瘤制药企业合作,将其作为癌症治疗中检测微小残留病的更可靠手段。”
除了无需进行内脏器官活检的便利性外,包骏还强调了该技术的成本优势。他说:“检测成本非常低,仅需约500元人民币,折合不到100美元。”
公司概况:EpiciphAI
总部地点:中国成都、北京
核心技术:通用型疾病液体活检技术,基于健康细胞与病变细胞拥有独特表观遗传“指纹”的核心原理
技术来源:北京大学
疾病研究方向:多类疾病(诊断技术不针对特定疾病)
临床阶段:发现阶段
成立时间:2025年,由何爱彬与Apuri生物风险投资公司联合创立
学术合作方:北京大学
企业合作方:暂无
员工数量:3人
融资计划:计划完成420万美元种子轮融资
投资方:未披露
首席执行官:包骏
专利情况:暂无已授权专利
2026 生物世纪公司 保留所有权利——仅供个人使用