The extraordinary diversity of protein sequences and structures gives rise to a vast protein functional universe with extensive biotechnological potential. Nevertheless, this universe remains largely unexplored, constrained by the limitations of natural evolution and conventional protein engineering. Substantial evidence further indicates that the known natural fold space is approaching saturation, with novel folds rarely emerging. AI-driven de novo protein design is overcoming these constraints by enabling the computational creation of proteins with customized folds and functions. This review systematically surveys the rapidly advancing field of AI-based de novo protein design, reviewing current methodologies and examining how cutting-edge computational frameworks accelerate discovery through three complementary vectors: (1) exploring novel folds and topologies; (2) designing functional sites de novo; (3) exploring sequence–structure–function landscapes. We highlight key applications across therapeutic, catalytic, and synthetic biology and discuss the persistent challenges. By fusing recent progress and the existing limitations, this review outlines how AI is not only accelerating the exploration of the protein functional universe but also fundamentally expanding the possibilities within protein engineering, paving the way for bespoke biomolecules with tailored functionalities.

Photoreceptor proteins utilise chromophores to sense light and trigger a biological response. The discovery that adenosylcobalamin (or coenzyme B₁₂) can act as a light-sensing chromophore heralded a new field of B₁₂-photobiology. Although microbial genome analysis indicates that photoactive B₁₂-binding domains form part of more complex protein architectures, regulating a range of molecular–cellular functions in response to light, experimental evidence is lacking. Here we identify and characterise a sub-family of multi-centre photoreceptors, termed photocobilins, that use B₁₂ and biliverdin (BV) to sense light across the visible spectrum. Crystal structures reveal close juxtaposition of the B₁₂ and BV chromophores, an arrangement that facilitates optical coupling. Light-triggered conversion of the B₁₂ affects quaternary structure, in turn leading to light-activation of associated enzyme domains. The apparent widespread nature of photocobilins implies involvement in light regulation of a wider array of biochemical processes, and thus expands the scope for B₁₂ photobiology. Their characterisation provides inspiration for the design of broad-spectrum optogenetic tools and next generation bio-photocatalysts.

Optogenetics has substantially enhanced our understanding of biological processes by enabling high-precision tracking and manipulation of individual cells. It relies on photosensitive proteins to monitor and control cellular activities, thereby paving the way for significant advancements in complex system research. Photosensitive proteins play a vital role in the development of optogenetics, facilitating the establishment of cutting-edge methods. Recent breakthroughs in protein design have opened up opportunities to develop protein-based tools that can precisely manipulate and monitor cellular activities. These advancements will significantly accelerate the development and application of optogenetic tools. This article emphasizes the pivotal role of protein design in the development of optogenetic tools, offering insights into potential future directions. We begin by providing an introduction to the historical development and fundamental principles of optogenetics, followed by an exploration of the operational mechanisms of key photosensitive domains, which includes clarifying the conformational changes they undergo in response to light, such as allosteric modulation and dimerization processes. Building on this foundation, we reveal the development of protein design tools that will enable the creation of even more sophisticated optogenetic techniques.

Accurate prediction of protein function is fundamental to progress in biotechnology and biomedicine, yet progress remains severely hampered by the widening chasm between exponentially growing genomic data and the limited capacity for functional annotation. High-throughput sequencing and metagenomics have driven an explosion in sequence data that far outstrips experimental characterization. UniProt now contains over 203 million protein entries, of which only ~2% have been experimentally validated. This widening “sequence–function gap” exceeds the reach of traditional homology-based tools such as BLAST (v2.17.0) and HMMER (v3.2), which are inherently constrained by sequence identity thresholds. The emergence of Protein Language Models (PLMs), including ESM and ProtTrans, has introduced a transformative paradigm, thereby shifting functional inference from similarity-based retrieval to geometric reasoning within learned semantic spaces. Nevertheless, current approaches remain largely confined to unimodal or narrowly bimodal frameworks, failing to capture the inherently multidimensional determinants of enzymatic function, including active-site geometry, chemical reaction logic, and literature-embedded semantic context. This review systematically adopts a multimodal global-fusion perspective, elucidating how three-dimensional geometric features, chemical reaction semantics, and textual knowledge graphs are synergistically integrated around PLMs as a core backbone. We delineate complementary mechanisms and integration strategies that together enable fine-grained protein function annotation beyond the performance ceiling of single-sequence methods. Furthermore, we survey the translational potential of such frameworks from computational prediction to real biological applications, and critically examine persistent bottlenecks including activity cliffs, transition-state inference, and conformational dynamics. We identify the integration of physics-informed machine learning with dynamics-aware architectures as a pivotal direction toward a causal, mechanism-level understanding of protein function.

Photoreceptors control cellular processes in response to light. Most photoreceptors sense blue or red light, but the recent discovery of the cobalamin‐dependent photoreceptor, CarH, has expanded the wavelength range of photoreception to other regions of the electromagnetic spectrum to include the green light region. Further identification of cobalamin‐dependent green light‐sensitive photoreceptors has been hampered owing to poor annotation of the light responsiveness of cobalamin‐binding domains (CBDs) in public databases. Here we report a computational workflow, SignatureFinder, that uses a combination of sequence and structural analyses to identify new light‐responsive CBD‐containing proteins. The light response of exemplar proteins containing the proposed signature were confirmed experimentally. A structural analysis of these new photoreceptors, including the crystal structure of a new CBD domain, highlights how the signature elements interact with the cobalamin chromophore to sense light. Database mining of 128 000 CBD‐containing sequences using the identified signature revealed more diverse CBD‐containing photoreceptors, thereby expanding the family of green‐light photoreceptors. A SignatureFinder web server is available (https://enzymeevolver.com) for wider applications, including the identification of signature sequences of other biological ligands of interest.