Protein Science Research: Decoding the Basic Language of Life

If we compare life to a finely tuned machine, DNA is the blueprint, RNA is the instruction manual, and proteins are the actual working "parts" of this machine—from enzymes that catalyze metabolic reactions, to antibodies that fight pathogens, from signaling molecules like cytokines, to structural proteins like collagen. Almost all life functions are carried out by proteins.

The science that studies proteins is called Protein Science. It’s one of the core foundational disciplines of life sciences and a crucial bridge connecting basic biology with the biomedicine industry. From understanding what a protein looks like, to figuring out what it does, how it’s regulated, and how to modify it—every step forward in protein science drives advancements in biomedicine, synthetic biology, green manufacturing, and other industries.

I. What does protein science study? Five core directions

The Five Core Research Areas in Protein Science

Protein science is not a single discipline but a vast interdisciplinary research system. It spans multiple dimensions, from molecules to cells, from structure to function, and from natural proteins to artificially designed ones. Its core can be summarized into five main areas.

1. Structural Biology: Analyzing proteins at the atomic-level 3D structure

Protein function is determined by its own 3D folded structure. Physiological roles such as molecular binding and catalytic activity all depend on specific spatial configurations. High-resolution structural analysis is a prerequisite for studying molecular mechanisms and developing targeted drugs.

Common techniques include X-ray crystallography, Cryo-EM, and NMR. In recent years, Cryo-EM has undergone technological innovations, making it possible to achieve near-atomic resolution structural analysis for challenging samples like membrane proteins and large protein complexes. AI protein structure prediction models have further lowered experimental barriers, allowing most proteins to obtain 3D structural models close to experimental accuracy without years of laborious lab work.

2. Proteomics: Globally and dynamically analyzing cellular protein expression profiles

The proteome is defined as the full set of proteins expressed in a specific cell, tissue, or organism under a certain physiological or pathological state. Proteomics relies on high-throughput technologies to systematically detect protein expression levels, modification states, molecular interactions, and functional changes.

While the genome has static properties, protein expression varies dynamically across time and space. The protein profiles of the same genome can differ significantly between cells, developmental stages, and disease conditions. Comparing protein expression differences between healthy and diseased samples can help identify disease biomarkers, uncover new drug targets, and clarify pathogenic molecular mechanisms.

According to the 2026 Global Market Insights report on the proteomics industry, the global proteomics market reached $41.2 billion in 2025. It is expected to grow from $45.7 billion in 2026 to $126.3 billion in 2035, with a compound annual growth rate of about 12%. The rapid growth reflects the strong demand for in-depth protein-level understanding in precision medicine and drug development.

3. Protein Function and Molecular Interactions: Clarifying protein pathways and interaction networks

This area focuses on protein catalytic activity, intracellular signal transduction, and protein-protein/protein-nucleic acid/protein-small molecule binding mechanisms, determining subcellular localization and the biological pathways involved.

Proteins rarely perform physiological functions independently; they rely on complex interaction networks to collaboratively regulate cellular activities. Systematic studies of these interaction networks, also called interactomics, are core methods for unraveling molecular life mechanisms.

4. Protein Engineering and De Novo Design: Customizing and Creating Artificial Proteins

This field has outstanding application value, aiming to break through the performance limits of natural proteins by modifying them or designing new proteins from scratch with desired traits. Traditional techniques include site-directed mutagenesis, directed evolution, and domain recombination, which heavily rely on manual screening and involve long development cycles and high costs.

AI technology has completely reshaped the R&D model, covering the full process from structure prediction, functional optimization, to de novo protein design, transforming protein engineering from trial-and-error experiments into standardized, quantitatively predictable engineering. The 2024 Nobel Prize in Chemistry was awarded to David Baker, Minkyung Baek in the field of protein design, and Demis Hassabis and John Jumper in protein structure prediction, officially marking the highest scientific recognition for this area.

5. Post-Translational Modifications (PTM): Decoding the Chemical Tags that Regulate Protein Activity

Polypeptides synthesized by ribosomes do not have full functionality yet; they need to fold into the correct 3D structures, bind cofactors, undergo post-translational modifications like phosphorylation/glycosylation/acetylation/ubiquitination, and assemble into functional complexes to gain complete physiological activity. So far, over 400 types of modifications have been identified. Each modification acts like a chemical tag, precisely controlling protein stability, intracellular localization, and molecular interactions.

Identical amino acid sequences can perform entirely different physiological functions depending on specific modifications. Systematic study of these modifications is called post-translational modification proteomics, which is a core research area for understanding cellular signaling regulation and discovering new drug targets.

II: Why is Protein Science So Important?

Protein science is a central discipline in life sciences. It connects upward to genomic and transcriptomic information and downward to cell physiology and organismal development research. It also extends into biomedicine, synthetic biology, and green manufacturing industries. Its value is reflected in three main aspects:

1. Basic Science Level: Understanding the Molecular Basis of Life

What is life? On a molecular level, life is a dynamic process where proteins, nucleic acids, metabolites, and other biomolecules interact in an organized manner across time and space, with proteins serving as the main carriers executing almost all core physiological functions. Understanding protein structure, function, and regulation is fundamental to knowing how cells work, how organisms develop, and how diseases occur. Without advances in protein science, modern life sciences would lack essential support.

2. Biomedical Industry Level: The Lifeblood of New Drug Development

Almost all biopharmaceuticals—antibody drugs, fusion proteins, enzyme replacement therapies, vaccine antigens—are essentially proteins or protein complexes. Most small-molecule drug targets are also proteins, such as enzymes, receptors, or ion channels.

Progress in protein science directly determines the efficiency and success rate of new drug development: identifying a good target protein can open up an entirely new therapeutic area; resolving the high-resolution structure of a target can significantly accelerate drug design and optimization; understanding the drug-target interaction mechanism can guide safer and more effective clinical use.

3. Green Industry Level: Synthetic Biology as the Core Engine of Industrial Manufacturing

The core of synthetic biology is using engineering methods to design and modify biological systems so they produce the substances humans need. And the "main players" in bio-manufacturing—enzymes—are basically proteins.

From starch enzymes and proteases for food processing, to cellulases for textiles, to lipases for bioenergy, and even new enzymes that can break down plastics—any breakthrough in industrial enzyme performance can potentially restructure an entire industry path. According to statistics from several industry research institutions, the global industrial enzyme market is expected to reach about $7.9–8.3 billion by 2025 and surpass $10 billion by 2030. The industry is steadily growing, and the core driving force behind this is the continuous progress in protein science and protein engineering.

III. What’s happening in protein science research in the AI wave?

If the core breakthroughs in protein science over the past twenty years were "seeing more clearly" (progress in structural analysis) and "seeing more" (advances in proteomics), then the most important change now is that AI is shifting protein research from "discovery-driven" to "design-driven."

This change shows up on three levels:

1. From "structure prediction" to "function prediction"

The first milestone of AI in the protein field was structure prediction. Today, structure prediction has become a standard tool—most protein structures can now be obtained through AI models with near-experimental accuracy.

The frontier is expanding into function prediction. Protein language models (PLMs), trained on billions of protein sequences, have learned the "grammar rules" of amino acid sequences, allowing direct prediction of protein functional properties from the sequence—stability, solubility, enzyme activity, affinity, immunogenicity… Properties that once required experiments can now be evaluated by AI in seconds, greatly reducing experimental work.

2. From "optimization and modification" to "de novo design"

Traditional protein engineering modifies natural proteins—changing a few amino acids to improve stability or activity. AI-driven de novo design starts directly from the functional need, creating entirely new proteins that don’t exist in nature.

This is a qualitative leap. It means humans are no longer limited by proteins evolved naturally and can "customize" functional molecules according to actual needs—like designing industrial enzymes that work under extreme conditions, designing entirely new antibodies targeting difficult drug targets, or creating artificial enzymes with new catalytic functions.

Since 2025, this direction has advanced especially fast. Research teams have de novo designed the most active artificial enzymes so far, and multiple teams have used AI to create highly stable artificial proteins that can tolerate temperatures over 100°C. Achievements like these were unimaginable ten years ago.

3. From "tool-assisted" to "full-process intelligence"

The deeper change lies in the research paradigm itself.

The traditional protein research process goes like this: researchers propose a hypothesis → design experiments → perform manual operations → analyze results → revise the hypothesis. It’s long, inefficient, and highly dependent on personal experience.

AI-driven new research paradigm, on the other hand, is an intelligent closed loop of "design—verify—iterate": AI handles sequence design, property prediction, and plan formulation; automated experimental platforms handle synthesis, expression, purification, and testing; experimental results are automatically fed back to the AI, which then optimizes the next round of design. The efficiency and success rate of the whole process far exceed the traditional model.

This closed-loop system is turning protein research from "lab craftsmanship" into an "engineered process."

IV. Practical case in China: MatwingsVenus™ (Xiaowu™) platform’s new intelligent R&D paradigm

AI Protein Generation Model Workflow

In the wave of AI-driven transformation in global protein science research, Chinese companies are quickly catching up and developing their own unique approaches. Represented by Shanghai Matwings Technology's MatwingsVenus™ (Xiaowu™) platform, domestic companies are exploring a complete 'AI protein research and automated experiment' pathway.

The MatwingsVenus™ (Xiaowu™) platform positions itself as a conversational AI agent for protein R&D. Its core idea is not to create a single tool, but to build an intelligent workflow platform that covers the entire protein research process. This is reflected in three main aspects:

First, conversational interaction lowers the usage barrier. Researchers can describe their needs in natural language — for example, 'I want to improve this protein's thermal stability' or 'Help me design a binding protein targeting this site' — and the AI agent automatically breaks down the task, calls the appropriate tools, and generates a complete design plan. This allows lab personnel without a background in computational biology to leverage AI for protein research.

Second, an integrated toolchain covers the full process. The platform integrates over 200 specialized protein design and analysis tools, along with extensive functional annotation databases — from structure prediction, sequence design, and property evaluation to mutation optimization, druggability analysis, and codon optimization. Common computational tasks in protein research can all be completed on a single platform, removing the need to switch between different software.

Third, a closed loop between wet and dry experiments ensures practical results. More importantly, the MatwingsVenus™ (Xiaowu™) platform connects cloud-based AI computing with automated high-throughput lab equipment, allowing target sequences generated by AI to directly drive automated experiments for sample preparation and function testing. Experimental data then feed back to iterate models, continuously optimizing AI design capabilities.

The platform has already validated this approach in multiple areas: in immune regulatory receptor target design projects, dozens of AI-generated binding molecules showed clear cellular blocking activity in vitro; and after iterative redesign of the sweet protein Monellin, multiple mutants significantly increased sweetness while maintaining high thermal stability. These real-world cases prove that AI-driven protein R&D systems have moved from theoretical concepts to actionable, efficiency-enhancing standardized tools.

V. Outlook: The Next Decade of Protein Science

Over the past thirty years, protein science has gone through three major paradigm shifts: from studying individual proteins to panoramic proteome analysis; from low-resolution observations to atomic-level 3D structural analysis; and from using natural proteins to artificial modification and creating entirely new proteins. In the next ten years, the industry will continue to develop along three clear directions, while driving upgrades in biomedicine and biomanufacturing.

1. Continuous breakthroughs in spatial proteomics, achieving subcellular-level precise localization

Spatial proteomics is the fastest-growing segment in the proteome field. Research is no longer limited to "which proteins are in a cell" but focuses on precisely locating proteins within tissues, cells, and organelles. This provides a new dimension for studying the mechanisms of tumors and neurological diseases and supports the development of precise diagnostic reagents.

2. Dynamic analysis of protein interaction networks, expanding functional research depth

The industry focus will shift from static protein maps to tracking dynamic physiological processes, connecting single molecules, multi-protein complexes, and multi-level cellular pathway regulation networks. This will fully clarify how protein networks change dynamically during disease processes and uncover more new druggable targets.

3. Maturation of AI-driven protein design technology, leading to widespread industrial application

AI-driven de novo protein design will move from "technically feasible" to "efficient and stable mass production," with the success rate of design, druggability, and functional complexity continuously improving. Once protein design becomes standardized and controllable, biomedicine, industrial enzyme manufacturing, and synthetic biology will experience a new wave of industrial growth.

For China's life sciences industry, the integration and upgrade of protein science with AI is a key lever for achieving self-sufficiency in the industry chain. High-end research equipment, core biological reagents, therapeutic proteins, and specialized industrial enzymes can all leverage domestic AI protein research platforms to achieve technological breakthroughs, reshaping the global industrial competition landscape.