总部位于巴黎的初创公司 Rivercell 近日完成了一轮 2500 万美元的种子轮融资，用于构建一个能够预测人类细胞对药物及基因变化反应的人工智能模型。该公司于周三表示，本轮融资由 HV 领投，HCVC、Alven 和 Bpifrance Digital Venture 参与跟投。
Paris start-up Rivercell has raised a $25M seed round to build an AI model that predicts how human cells respond to drugs and genetic changes. HV led the round, with HCVC, Alven and Bpifrance Digital Venture taking part, the company said on Wednesday.
这笔资金将用于支持 Rivercell 的数据平台建设、扩大其位于巴黎的湿实验室规模，并推出其人工智能虚拟细胞（AI Virtual Cell）计划。该公司将该模型描述为细胞的世界模型，即能够预测细胞在接受治疗后将如何发生变化的模拟系统。
The money will pay for Rivercell’s data platform, a bigger wet lab in Paris and the launch of its AI Virtual Cell programme. The company describes the model as a world model of the cell, meaning a simulation that predicts how a cell will change after a treatment.
Rivercell 的联合创始人兼首席执行官扬·弗勒罗（Yann Fleureau）表示：“细胞的世界模型是医学领域百年一遇的机遇。而目前缺失的关键拼图就是数据。”
“A world model of the cell is the opportunity of the century in medicine. The missing piece is data,” said Yann Fleureau, Rivercell’s co-founder and chief executive.
为什么要构建自己的数据 Rivercell 表示，AlphaFold 之所以能够预测蛋白质结构，是因为它利用了数十年的公开数据。然而，该公司指出，目前并不存在同等规模的、关于细胞随时间对药物反应的数据集。因此，该公司计划自行生成这些数据。
Why it is building its own data AlphaFold could predict protein structures because it drew on decades of public data, Rivercell said. No dataset of that scale exists for how cells respond to drugs over time, according to the company. So it plans to generate its own.
其宣言中列出了现有数据的四个问题。其中大部分数据记录的是相关性，而非对干预措施的测试。它们通常只在单一时间点、针对极少数细胞类型，测量细胞的某一个方面。
Its manifesto lists four problems with existing data. Much of it records associations rather than testing interventions. It often measures one aspect of a cell, at a single moment, in too few cell types.
Rivercell 计划在每次实验过程中拍摄活细胞，然后读取其基因活性。其人工智能模型将选择下一个实验，并由实验室进行操作。该公司正在为此制造自己的仪器，并将其称为“生物数据中心的 GPU”。Rivercell 表示，该模型应当能够减少所需的实体实验室实验数量。
Rivercell plans to film living cells through each experiment and then read their gene activity. Its model will choose the next experiment, and the lab will run it. The company is building its own instruments for this, which it calls the “GPU for the bio data center”. Rivercell says the model should cut the number of physical lab experiments needed.
幕后团队 弗勒罗于 2025 年夏季创立了 Rivercell。在此之前，他是人工智能心脏诊断公司 Cardiologs 的联合创始人兼首席执行官，该公司于 2021 年被飞利浦收购。
Who is behind it Fleureau started Rivercell in summer 2025. He previously co-founded Cardiologs, an AI heart diagnostics company that Philips bought in 2021.
埃里克·杜兰德（Eric Durand）于 2026 年加入该公司，担任联合创始人兼首席科学官。他曾在诺华（Novartis）领导肿瘤学数据科学工作长达七年，随后在 Owkin 负责数据科学业务，并联合创立了 Bioptimus。
Eric Durand joined in 2026 as co-founder and chief scientific officer. He spent seven years leading oncology data science at Novartis, then ran data science at Owkin and co-founded Bioptimus.
EPFL的Charlotte Bunne和慕尼黑亥姆霍兹研究所的Fabian Theis担任科学顾问委员会成员。天使投资人包括Twist Bioscience、DNA Script和Agilent的创始人及高管。
Charlotte Bunne of EPFL and Fabian Theis of Helmholtz Munich sit on the scientific advisory board. Angel investors include founders and executives from Twist Bioscience, DNA Script and Agilent.
“AI虚拟细胞将彻底改变我们对生物学的理解和改造方式，”EPFL助理教授Charlotte Bunne表示。
“AI virtual cells will transform how we understand and engineer biology,” said Charlotte Bunne, an assistant professor at EPFL.
其他细胞和药物模型 其他公司也在构建细胞的AI模型。伦敦的Relation公司正与GSK合作开发细胞基础模型。在AI药物发现领域，Chai Discovery于7月完成了4亿美元的C轮融资，而字节跳动的Anew Labs则在9月获得了2.9亿美元融资。
Other cell and drug models Other companies are also building AI models of cells. London’s Relation is working on a cell foundation model with GSK. In AI drug discovery, Chai Discovery raised a $400M Series C in July, and ByteDance’s Anew Labs raised $290M in September.
Rivercell并未透露其估值，也未说明首个模型预计何时完成。
Rivercell did not give a valuation or say when it expects its first model to be ready.