BlueRun Ventures Headlines | BioMap's First Pipeline Shows Promising Preclinical Data, Launches Immune Therapy Toolbox

What happens when you put AI models to work in drug development? As the world's largest life-sciences foundation model moves into pipeline and platform validation, could plug-and-play, modular, prefabricated therapeutics driven by large models become the future of "tailor-made" precision medicine? If these questions intrigue you, today's article shares the explorations and progress of BlueRun Ventures portfolio company BioMap —

By combining cutting-edge AI with biotechnology, BioMap has built xTrimo, a 100-billion-parameter life-sciences foundation model centered on protein language. On this foundation, it developed AIGP, an AI generative protein design platform. By modeling biological evolution and decoding the laws of life, BioMap has achieved breakthrough results across disease target discovery, innovative antibody drug design, and peptide and enzyme engineering.

In 2023, BioMap reached a commercial collaboration with Sanofi with total deal value exceeding $1 billion for AI model development, pioneering milestone payments tied to model development progress rather than drug R&D milestones — a novel case of the MaaS (Model as a Service) business model in the large-model era.

But how effective are these AI models in practice, and what capability leaps can they bring to drug development? After two years of refinement, BioMap's innovative drug R&D portfolio ImmuBot has also delivered a demonstration: first-batch pipeline preclinical data has met expectations and is steadily advancing toward clinical stages. Additionally, BioMap has developed a proprietary plug-and-play modular immune-cell engager multispecific antibody platform capable of precise, generalizable combinations across multiple immune cell types and tumor targets, opening new possibilities for immunotherapy.

Over the past decade, immunotherapy has fully activated key markets in oncology and autoimmune disease. Yet the pursuit of more precise, more efficient, better-balanced broad-spectrum and targeted precision medicine has never ceased. As the world's largest life-sciences foundation model enters pipeline and platform validation, could large-model-driven plug-and-play, modular, prefabricated therapeutics become the future of "tailor-made" precision drugs? VB Data (动脉网) sat down with Dr. Zhaoshi Jiang, Senior Vice President of BioMap, to explore these questions.

Dr. Zhaoshi Jiang

Senior Vice President, BioMap

Dr. Zhaoshi Jiang brings 20 years of experience in bioinformatics, drug target discovery, and drug design. Before joining BioMap, he served on Gilead Sciences' drug discovery leadership team as Executive Director and Head of Bioinformatics and Data Science. His team supported nearly 100 clinical trials and contributed to the approval of nearly ten new drugs, including notable therapies such as Jyseleca (autoimmune), Vemlidy (hepatitis B), Biktarvy (HIV), and Remdesivir (viral infections). Beyond clinical R&D, he has made significant contributions to genomics research, publishing over 40 high-quality scientific papers including five in Nature. Prior to Gilead, he was a senior scientist at Genentech, leading multiple large-scale cancer genomics initiatives. He earned his Ph.D. in genomics from the University of Washington under Professor Evan Eichler, a member of the U.S. National Academy of Sciences and renowned geneticist. He previously graduated from Zhejiang University's clinical medicine program and worked for many years as a clinical pathologist at Peking Union Medical College Hospital.

First-Batch Pipeline Preclinical Data Meets Expected Targets

Building a Plug-and-Play Immunotherapy Toolbox

VB Data: Since its debut in September 2022, what internal iterations and new progress has ImmuBot undergone?

Jiang: As an AI-centric company, BioMap began by collecting unprecedented, rich datasets — downloading, cleaning, and integrating nearly all compliant public multi-omics big data related to human immunology — to build the world's largest life-sciences multimodal pre-trained model.

Based on a high-throughput wet-dry closed-loop bio-computing engine to parse the immune system, we use large models to precisely predict and decode complex immune patterns and disease mechanisms, with the aim of developing next-generation protein or antibody drugs to achieve precise treatment of currently unmet immune-related diseases. It was in September 2022 that we first disclosed ImmuBot, BioMap's breakthrough innovative drug portfolio.

To date, we are pleased to share that based on the company's distinctive technology, we have successfully completed the design, optimization, construction, and preclinical validation of the ImmuBot innovative drug asset portfolio, which includes the following functional features:

a. High-performance immune-cell engager building blocks targeting effector T cells, natural killer (NK) cells, gamma-delta T cells, and various other immune cell types;

b. Innovative tumor-associated antigen targets covering diverse solid tumors. These targets are self-developed based on BioMap's bio-computing platform's powerful data foundation and unique target recommendation algorithm technology, with some being first-in-class. They serve as high-precision tumor-targeting warheads, becoming key building blocks for next-generation bispecific antibodies or ADC drugs;

c. Functional trigger components for complex immune microenvironments, which are expected to further improve the therapeutic window of next-generation drugs;

d. The asset portfolio features plug-and-play modular design, allowing flexible mobilization and rapid assembly of multiple immune regulatory elements according to patient immune microenvironment heterogeneity and the component needs of different indications, achieving reprogramming of the immune system to precisely identify and kill tumors in a highly efficient, low-toxicity manner, with the expectation of bringing revolutionary breakthroughs to tumor immunotherapy;

Based on these drug components and platforms, the company has developed multiple bispecific antibody drug pipelines centered on liver cancer, gastrointestinal solid tumors, and autoimmune diseases, particularly for patients who are resistant to or respond poorly to current first-line immunotherapies. Some pipelines have reached preclinical stages, with in vitro and animal experiments demonstrating very good efficacy and safety in our designs, meeting expected design targets, and are planned for further advancement to CMC and clinical development.

VB Data: At a time when AI innovations are breaking through biotechnology bottlenecks, how does the ImmuBot immunotherapy platform pioneer the path toward precision immunotherapy?

Jiang: To address the pain points in immunotherapy and related drug R&D and further enhance precision, our ImmuBot drug platform is designed to achieve the following goals:

First, building a refined, regulatable comprehensive immunotherapy toolbox.

To start with, refined and regulatable. Taking our first core bispecific antibody pipeline as an example, this is a conditionally activated CD3 T-cell engager bispecific antibody whose unique molecular design ensures the drug remains in "standby" mode in non-tumor tissue, only being activated by specific proteases in the tumor microenvironment to perform targeted tumor cell killing. This conditional trigger mechanism is expected to greatly reduce on-target/off-tumor toxicity risks of CD3 antibodies and expand the drug's therapeutic window. In comparative experiments, we have observed activation efficiency and therapeutic window several-fold higher than industry competitors, and have seen very positive tumor suppression effects and high safety in in vivo animal experiments.

Next, a comprehensive immunotherapy toolbox: Beyond conditional activation, we also have logic gates, enhancers, and various other functional regulatory elements, giving our immunotherapies powerful precise targeting and precise activation capabilities. Beyond CD3 T cells, we have also deployed natural killer (NK) cell, gamma-delta T cell, and various other immune-cell engagers, enhancing their killing efficacy through various regulatory elements. These innate immune-cell engagers, with their relatively higher safety profiles, are expected to show tremendous potential in future tumor combination therapies.

Second, establishing a library of highly specific tumor-targeting navigation warheads.

As an AI platform company in the life sciences, BioMap possesses strong database resources, target discovery algorithm capabilities, and experimental validation closed-loop capabilities. Combining a 100-billion-level public immune database with over 100 million unique proprietary experimental data points, we have established a top-tier immune target discovery platform and identified a series of globally first-in-class novel targets.

Many existing druggable targets are also expressed at low levels in normal tissue, bringing off-tumor toxicity risk. Therefore, we have combined target differential expression and subcellular localization prediction to discover new targets inaccessible to traditional methods, and for difficult-to-drug targets with toxicity risks, we have applied logic gate designs to further enhance safety. We believe these targets and target combinations will shine as directional warheads for next-generation immune-engager bispecifics or ADC drugs.

Modular Drug Design:

Finding the Balance Between Personalized Medicine and Scalable Manufacturing

VB Data: What pain points in immunotherapy can the ImmuBot immunotherapy platform and innovative protein drugs address?

Jiang: I believe immunotherapy is the most critical direction for treating human complex diseases. Taking tumors as an example, why do tumors have certain characteristics of aging? Simply put, the immune system is a surveillance system with strong monitoring of cancerous tumor cells. In young bodies, tumor cell mutation probability is low and the immune system is relatively strong, likely able to clear tumor cells. With aging, immune system function declines, carcinogen exposure time extends, and mutation possibility gradually increases — as one side weakens, the other strengthens, making tumors relatively more prevalent in elderly populations.

Therefore, from a treatment perspective, activating and enhancing the body's own immune mechanisms to kill, regulate, and treat tumors is the most natural and most efficient approach. This is also why immunotherapy can achieve excellent long-term tumor control in some patients.

Over the past 20 years, a major advance in oncology has been the invention of immune checkpoint inhibitors, which addressed the limitation of relying solely on radiotherapy, chemotherapy, ADC, and targeted therapy. However, the efficacy rate of current first-line immune checkpoint inhibitors still needs improvement, with only about 20-30% of patients responding on average. Some patients develop resistance or relapse after treatment. Additionally, first-generation T-cell-based immunotherapies have severe toxic side effects, limiting their applicability. In short, tumor immunotherapy still has enormous room for expansion.

Where do these problems originate? At their root, because every person is different. Through multi-omics studies of tumor microenvironment characteristics, especially single-cell sequencing and spatial transcriptomics, we find that tumor patient tissue microenvironments have strong heterogeneity. This heterogeneity manifests as: different infiltration levels of different immune cells, tumor benign/malignant degree, and diversity of tumor-associated mesenchymal stem cell components. These differences all affect the clinical efficacy of immunotherapy.

Therefore, future tumor treatment requires personalized design, developing treatment plans matched to patient characteristics to address the current challenge of variable immunotherapy efficacy. Designing products based on patient diversity requires sufficient product flexibility —

The ImmuBot design fully embodies this philosophy: modular, plug-and-play, flexible product building blocks, from tumor-associated targets to different immune engagers to functional regulatory elements, bring technical possibilities for future personalized designs for different diseases, different patients, and different clinical stages.

VB Data: Personalization is a major trend in precision medicine, but it also faces contentious issues around industrialization, scalability, accessibility, and cost. How do you view this debate?

Jiang: The personalization of precision medicine certainly involves challenges in industrialization and scalable production. However, personalization does not mean the ideal state of designing a product for each individual patient. Our philosophy is finding a balance between personalized design and scalable production.

Traditional treatment plans using the same drugs at the same doses for all patients are clearly unreasonable. Because each patient's tissue microenvironment differs, variable clinical efficacy and toxic side effects are not surprising. But the challenges of personalization are: first, clinical R&D challenges. In regulatory review, finding balance between personalized design and providing sufficient evidence of efficacy and safety is not easy. Second, manufacturing costs would be very high. Personalized design and scalable product production are not easily reconciled.

This is why BioMap emphasizes the plug-and-play concept today: we believe not every patient needs a unique module and product. Taking a certain tumor type as an example, setting certain biomarkers as identification features, patients can be divided into multiple possible subpopulations. Within patient subpopulations, designing one or multiple combined or sequential treatment regimens respectively, we can hopefully achieve the optimal balance point between personalized treatment and scalable production.

VB Data: So what differentiated advantages do prefabricated, flexibly assembled, plug-and-play, modular characteristics bring to the ImmuBot immunotherapy discovery platform?

Jiang: Our immune-cell engager multispecific antibody platforms have all achieved highly flexible plug-and-play characteristics at the antibody design level, meaning that when developing "immune robot" drug molecules with different characteristics, various functional building blocks can be flexibly deployed and rapidly assembled according to needs (such as for different tumors and associated antigens), ensuring drug molecules achieve optimal performance in low-toxicity, high-efficacy states while reducing development costs and timelines.

The various functional building blocks we design have reuse potential, bringing tens or even hundreds of times improvement to overall drug R&D efficiency, and minimizing challenges that new products may bring in downstream scalable production.

Most importantly, based on this technology, we will be able to develop more "tailor-made" precision drugs for various finely divided patient subpopulations. Furthermore, this also facilitates R&D of drugs with greater market and clinical urgency.

VB Data: Being both platform and drug (pipeline), how does BioMap envision ImmuBot's commercialization model?

Jiang: For BioMap, the primary business model will remain focused on external services for AI models and platforms in the life sciences. ImmuBot serves as an important application scenario and demonstration project in our biopharmaceutical development. We hope to further advance these high-value drug components and pipelines toward the clinic.

Based on immune regulatory building blocks and highly specific tumor targets, combined with deep exploration of unmet clinical needs, we have independently initiated or collaboratively developed multiple bispecific antibody drugs that are progressively completing lead confirmation and in vivo validation, with best-in-class potential already emerging. We very much look forward to, with the support of partners and investors, rapidly advancing these excellent molecules and drug pipelines into clinical stages, ultimately benefiting patients.

Meanwhile, the progressive validation of these drug pipelines also means ImmuBot's comprehensive immune-cell engager platform, immune regulatory building block platform, and target combination discovery platform have been firmly established and formed. Going forward, we warmly welcome interested partners to join hands with us in exploring the infinite possibilities of precision immunotherapy.

AI Prediction Will Empower Refined Combination Strategies in Immunotherapy

VB Data: You just mentioned designing one or multiple immunotherapy combination regimens around different populations or biomarker specificities. How do you view immunotherapy combinations?

Jiang: In the past, I worked on HIV therapy R&D at Gilead. AIDS was initially considered a fatal disease, until the proposal of cocktail therapy — using three or more drugs in combination to reduce viral resistance or escape from single-drug treatment, maximally suppressing viral replication, restoring damaged immune function, and extending patient survival. Today, in 10-20 year long-term HIV treatment, resistance basically does not occur; with one pill daily, patients can achieve long-term, high-quality survival, essentially transforming a terminal illness into a chronic disease.

This success story has demonstration effect for tumor treatment as well. Today's single therapies, whether chemotherapy, ADC, or immune checkpoint inhibitors, struggle to completely eradicate solid tumors. In the foreseeable future, effective combination or sequential treatment regimens are expected to bring tumor patients long-cycle (20-30 year), high-quality lifespans. If we can achieve this goal, in fundamental terms, this is already functional cure of tumors. This is also the fundamental reason we are building a plug-and-play, modular comprehensive immunotherapy toolbox today and laying out multiple immunotherapy pipelines.

VB Data: What directions can AI and large language models empower for combination therapies or multi-mechanism combination strategies?

Jiang: First direction: discovery of innovative target combinations. When single targets are insufficiently effective, AI algorithms and data science can help with what new targets patients should select, what biomarker expression levels, and what target combinations achieve optimal treatment efficacy and highest safety.

Second direction: AI may assist in designing complex individualized combination therapies and component selection. The complexity of selecting combination therapies far exceeds today's simple experience-based drug combination schemes. For example, biomarker expression levels may not closely relate to prognosis, and combinations of multiple immune phenotypes and tumor microenvironment characteristics affecting combination efficacy all need to be incorporated into combination therapy design considerations.

Third direction: how immunotherapies can precisely strike tumor tissue is essentially finding differential features from normal tissue. For example, ImmuBot designs incorporate conditionally triggered components such as tumor microenvironment enzyme profile changes, pH changes. Through sufficient conditional triggering, ImmuBot can further increase therapeutic window and enhance safety and efficacy.

VB Data: From the broader perspective of biological computing and AI computing, what is the significance of ImmuBot immune robot explorations?

Jiang: From its establishment, BioMap has been a platform company seeking to empower life sciences through AI and data science. Excellent AI companies can never be castles in the air or highbrow and unpopular; after having very advanced models, they must have grounded, applicable practical validation. If platform prediction capability is strong, then future molecular design success rates, product efficacy, and R&D timelines should all show clear improvement, directly demonstrating AI platform value.

Conversely, the large amounts of real data collected in drug development practice also bring more targeted data support for further optimizing large language models' prediction capabilities in drug design, helping models perform more refined fine-tuning and thereby improving model prediction capabilities. Thus the two actually have good interactive effects.

In the past, AI was relatively mature in small-molecule drug design, while protein drug design was a large blank space. Protein 3D structure prediction, protein affinity prediction, and others have long been major technical challenges on AI computing platforms. But today, a series of breakthroughs brought by high-performance computing and cutting-edge AI algorithms are increasingly surpassing traditional experimental method limitations.

We have seen cases where targets previously considered "difficult to drug" or "undruggable" are gradually being broken through with computational assistance. In the next five to ten years, with computer and AI assistance, innovative protein drug products will continuously emerge, completely transforming the landscape of this track and bringing benefits to patients.

---

BlueRun Ventures was established in 2005 and is a venture capital firm focused on early-stage startups.

Currently, BlueRun Ventures manages multiple USD and RMB dual-currency funds in China, with assets under management exceeding RMB 15 billion, making it one of the largest early-stage funds domestically. Its investment stage focuses on Pre-A and Series A rounds, covering technology, consumer, and healthcare sectors, with nearly 200 portfolio companies including Li Auto, Waterdrop, QingCloud, Guazi.com, Qudian, Ganji.com, Energy Monster, Gaussian Robotics, Songguo Mobility, Yuntusemiconductor, Machenike, Yunsheng Intelligence, Anxinnet Shield, BioMap, and others.

BlueRun Ventures has been ranked #1 in Zero2IPO Group's "China Early-Stage Investment Institutions Top 30," #1 in ChinaVenture's "China Best Early-Stage Venture Capital Institutions TOP30," and has been named among Preqin's Top 10 VC fund managers globally for sustained high-return performance.

Additionally, BlueRun Ventures has received consecutive annual awards from Forbes China, 36Kr, Cyzone, Caixin Media, CBNweekly, Jiemian, and other media institutions for honors including "China's Best Early-Stage Firm," "China's Top Venture Capital Firm," "Most Entrepreneurur-Friendly Early-Stage Firm," and "Most Influential Early-Stage Firm."