Transposon-associated TnpB is a compact and versatile gene editor that holds significant promise in the life sciences. However, current engineering strategies for TnpB are limited, resulting in low cleavage efficiency in mammalian cells that restricts its broader use. In this study, we developed a method termed TnpB-GMRE, which integrates a generative protein model for TnpB with a virtual screening pipeline based on the minimum recovery rate and energy minimization. From 100,000 generated sequences, we selected the top five candidates for experimental assessment. Without the use of enrichment strategies such as flow cytometry or antibiotic selection, three of the five mutants displayed higher editing activity across four target sites. Among them, the TnpB-TD mutant achieved the highest editing activity (17.7%), representing a 50% increase compared to the original ISDra2 TnpB from Deinococcus radiodurans. In addition, the TnpB-TD mutant exhibited greater diversity in editing profiles and superior efficiency in deleting long fragments (> 10 bp) relative to the intact ISDra2 TnpB. Molecular dynamics simulations revealed that compared to the wild type, the TnpB-TD mutant adopted a greater number of low-energy conformational states and displayed an increased positive charge on its surface, suggestive of a stabilized structure that may underlie its enhanced editing performance. This strategy provides a framework for optimizing compact nucleases and broadens the available toolkit for gene-editing applications.

Rapid yet high-information-content monitoring of cellular metabolism during fermentation is highly desirable for process optimization. While single-cell Raman spectroscopy can rapidly distinguish intracellular biopolymer classes, finer-resolution discrimination of monomeric units remains challenging. In this study, we demonstrated that the ramanome can classify polyhydroxyalkanoate (PHA)-producing Halomonas bluephagenesis cells that synthesize either polyhydroxybutyrate or poly(3-hydroxybutyrate-co-4-hydroxybutyrate) with 99.75% accuracy and simultaneously quantify total intracellular PHA and its constituent monomeric units (3-hydroxybutyrate and 4-hydroxybutyrate) at the single-cell level (median absolute deviation <3.8%). Such information enabled timely, data-driven selection of the optimal harvest time and revealed precursor competition between nucleic acid and PHA biosynthesis. Moreover, when monitoring PHA production in an industrial-scale 5000 l-fermenter, our method achieved a 12-min turnaround time, representing a more than100-fold acceleration over gas chromatography. Furthermore, tests on protein production by Saccharomyces cerevisiae and on lipid synthesis in Rhodococcus opacus supported its versatility. Thus, ramanomics is a valuable approach for process control and strain evaluation in biomanufacturing.

Addressing global food insecurity requires sustainable alternatives to traditional agriculture. Converting agricultural residues such as corn stover into artificial starch and single-cell protein (SCP) represents a promising solution, while industrial deployment remains hindered by low products yield and high costs associated with key enzymes like alpha-glucan phosphorylase (alpha GP). This study aimed to address these challenges through intelligent mining and engineering of alpha GP and systematic bioprocess parameters optimization. Five novel alpha GP candidates (NaGP, LbGP, CbGP, NtGP and CaGP) with a critical CAP domain and high predicted expression propensity were selected, and NtGP exhibited superior expression and starch synthetic activity. Examination of the CAP domain highlighted 9 mutation sites in NtGP, indicating their role in starch synthesis. Specifically, the mutation within the CAP domain was shown to affect starch synthesis, with the N563T mutant displaying a 1.23-fold improvement in activity relative to the wild-type enzyme. Concurrently, a thermotolerant Candida utilis chassis capable of growing at elevated temperatures and utilizing both glucose and xylose was used for SCP production in the optimized reaction system. The synergistic strategy nearly doubled the starch yield compared to previously reported techniques, produced 5.95 g/L starch and 5.79 g/L SCP from 40 g/L pretreated corn stover. This work presents a novel strategy to improve the efficiency of agricultural waste bioconversion, addressing critical challenges in sustainable food and feed production.

Microbial enzyme production and catalysis systems are crucial aspect of biotechnological research. However, building them from trustworthy published experimental data presents a major obstacle for both manual and automated techniques. Here, we introduce MEPAM (Microbial Enzyme Production and Catalytic Activity based on LLM), a question-answering system designed to accurately address inquiries related to enzyme production and catalytic reactions. Specifically, by training three machine learning models with >0.98 accuracy, we identified 11,068 high-quality, relevant articles from the Web of Science. Leveraging DeepSeek-V3 with zero-shot learning, we developed an ontology-driven knowledge representation that extracted 12,434 entities and 35,918 relations with 0.78 extraction accuracy and constructed a structured knowledge graph. Compared to few-shot learning and other machine learning methods, our framework achieved significantly higher extraction accuracy. Using this framework, we developed MEPAM based on retrieval-augmented generation and prompt engineering. Finally, using MEPAM, we extracted a comprehensive network involving the expression profiles, precise culture conditions, and substrate preferences for cellulase, demonstrating the strong utility of this tool. Compared with traditional LLMs, particularly GPT-4o, MEPAM exhibited superior performance, achieving significantly higher answer accuracy (0.86 vs. 0.52) and nearly eliminating hallucinations. MEPAM is available at http://180.76.108.212. This framework provides context-rich, verifiable insights, thus bridging predictive modeling with experimental validation to facilitate the exploration of microbial enzymatic systems.

Uterine glands synthesize and secrete compounds essential for embryonic and fetal development. They are critical for pregnancy, and their optimal function enhances prenatal survival and thus increases sow lifetime productivity. The gut microbe-uterus axis plays an important role in uterine development and function. Probiotics, especially Limosilactobacillus reuteri (L. reuteri), have been fed to improve female reproductive function, however, the beneficial effects remain unknow. The current study isolated one native L. reuteri (LRY15) from sow intestines that increased the number of uterine glands, and there elevated the protein expression of an important gene FOXA2. Moreover, LRY15 increased the expression of uterine genes related to angiogenesis, placental development, cell junction proteins, antioxidant enzymes, hormone metabolism, and transcriptional factors. Simultaneously, the protein levels of these factors (including blood vessel markers CD31 and VEGF, and cell junction proteins ZO-1 and occludin) were higher in LRY15 uteri compared to controls. The beneficial blood metabolites such as the tocopherol metabolite, antioxidant metabolites including retinol and the polyunsaturated fatty acids, and the levels of some cholic acids metabolites were elevated in LRY15 gilt blood (P < 0.05). Moreover, LRY15 improved cell-cell junction formation to enhance intestinal function. This is the first report to show that the native probiotic LRY15 can enhance uterine development and function; it is therefore a promising tool for increasing mammalian reproductive capacity in the future.

Abstract Background Korshinsk peashrub (Caragana korshinskii Kom.) is a valuable forage shrub for ruminants due to its abundant cell walls. However, its utilization is largely limited by the cross-linked structures of lignocellulose within its cell walls. White rot fungi possess the ability to degrade these resistant cross-linked structures, offering enormous potential to develop cost-effective biopretreatment processes of Korshinsk peashrub. Results Among the white rot fungi evaluated, Dichomitus squalens demonstrated superior efficacy in improving lignocellulose deconstruction and subsequent rumen fermentation of Korshinsk peashrub (P < 0.05). This fungus preferentially degrades lignin, hemicellulose, and pectin (P < 0.05), which corresponded to significantly improved enzymatic saccharification, ruminal degradability, and gas production (P < 0.05). Genomic analysis revealed that D. squalens possesses a comprehensive range of genes encoding ligninolytic enzyme. Elevated activities and expression levels of laccase, manganese peroxidase, esterase, glutathione S-transferase, versatile peroxidase, and hydrogen peroxide-generating enzymes aligned with the disruption of cross-linked structures and increased porosity of Korshinsk peashrub. Furthermore, the extracellular enzyme cocktail from D. squalens exhibited robust lignin‑degrading capability, corroborating its role in selective ligninolysis. Conclusions Pretreatment of Korshinsk peashrub with selective white rot fungi offers a practical approach to valorize this woody biomass as an alternative feedstock for ruminants.

ABSTRACT Fungal enzymes in glycoside hydrolase family 5 subfamily 5 (GH5_5) display notable catalytic diversity, efficiently degrading cellulose and sometimes mannan. However, the structural determinants and molecular mechanisms governing substrate preference in this enzyme family remain unclear. In this study, GH5_5 enzymes from fungi were systematically classified using profile-based sequence models and functionally characterized. Saturation mutagenesis combined with high-resolution crystal structure analysis of the bifunctional enzyme BsCel5B, exhibiting cellulase (CEL) activity of 941 ± 17 U/mg and mannanase (MAN) activity of 1,736 ± 34 U/mg, was employed to identify key residues controlling substrate specificity. Residue T100 in BsCel5B was identified as a major structural contributor associated with significant shifts in substrate preference. The T100V and T100N mutations resulted in 2.0-fold increases in MAN activity and 2.5-fold increases in CEL activity, respectively, generating bifunctional enzymes with enhanced substrate-specific activities. Similar substrate specificity trends were observed in several GH5_5 cellulase mutants. Their structural analysis indicated that substrate preference in fungal GH5_5 enzymes might be shaped by residual network-mediated alterations of the active-site geometry, with T100 acting as a second-shell regulatory element within a cooperative residue network. Together, these findings suggest a mechanistic framework for engineering catalytic specificity in GH5_5 enzymes.IMPORTANCECellulose and mannan are major components of plant biomass, and enzymes capable of efficiently breaking them down are essential for sustainable biofuel production and biomass utilization. Fungal enzymes in GH5_5 are widely used for these purposes, yet their functional diversity has been difficult to predict or control. Substrate preference in these enzymes can be modulated by altering a single amino acid, offering a promising approach for tuning enzyme activity. The identification of a key residue that influences the balance between cellulose and mannan degradation provides valuable insights for engineering enzymes with tailored functions. These findings contribute to a deeper understanding of fungal biomass-degrading enzymes and support the rational design of more efficient catalysts for industrial and environmental applications.

Extremozymes offer substantial potential as biocatalysts in industrial biotechnology, yet their identification and optimization remain challenging. Here, we developed AAEPre, a transfer learning-based predictor for acidophilic and alkalophilic proteins, trained on a curated non-redundant dataset. AAEPre achieved an average accuracy of 0.80 and outperformed conventional machine learning approaches. Based on this model, we developed an integrated pipeline for mining and engineering alkalophilic and thermophilic enzymes, combining sequence-based prediction, generative modeling, and multi-parameter virtual screening. This strategy enabled the discovery of a novel xylanase, 8E20, with optimal activity at 55°C and pH 8.0, followed by large-scale in silico diversification to generate 1000,000 variants. Systematic screening identified the superior variant 8E20-178, which exhibits a 1.9-fold increase in catalytic activity, a shift in optimal pH from 8.0 to 10.0, and improved alkaline stability. Structural analysis suggests that strengthened hydrophobic interactions and charge redistribution contribute to its improved alkali tolerance. Notably, 8E20-178 has strong potential for practical use, including pulp biobleaching and beating. The AAEPre model now is available at http://106.8.105.46:10152/, and is free for users. Collectively, our work presents a generalizable and experimentally validated computational framework for enzyme discovery and optimization under extreme conditions.

Cottonseed protein, a major agricultural by-product, represents a promising and sustainable resource for bioactive peptides discovery. In this study, four novel peptides (P1: EGPGCPMMER, P2: ETEDACR, P3: LLLNCKADK and P4: SSRFCTLPQQ) were isolated from cottonseed protein hydrolysate using multi-steps separation. All four peptides demonstrated dual antioxidant activities: directly scavenging Reactive Oxygen Species (ROS) and activating the Kelch-like ECH-associated protein 1/Nuclear factor erythroid 2-related factor 2 (Keap1/ Nrf2) signaling pathway, albeit with varying efficacy. Radical scavenging assays confirmed their potent activity, with 1,1-diphenyl-2-picrylhydrazyl (DPPH) EC50 values of 0.18 f 0.02, 0.45 f 0.05, 0.89 f 0.07, and 0.43 f 0.03 mg/mL for P1 to P4, respectively. EGPGCPMMER (P1) exhibited the most potent direct scavenging ability, with efficacy comparable to glutathione. In cell assays, pretreatment with all four peptides activated the Keap1/ Nrf2 pathway, as evidenced by promoted Nrf2 nuclear translocation in Hepatocellular carcinoma (HepG2) cells, leading to an enhanced cellular antioxidant defense system. While P1 excelled in direct scavenging, the P3 group demonstrated the most significant reduction in malondialdehyde (MDA) levels, indicating a superior capacity to mitigate lipid peroxidation. Molecular docking analysis revealed stable interactions between all four peptides and the Kelch domain of Keap1 for key residues such as Arg380, Arg415, and Tyr334. Molecular dynamics simulations further demonstrated that all four peptides formed stable complexes with Keap1. Overall, these newly identified cottonseed-derived peptides exhibit complementary antioxidant functions, underscoring their potential as natural agents against oxidative stress in various applications.

The low efficiency of enzymatic saccharification of chitin severely limits its potential for high-value applications. Consequently, developing an understanding of the chitinolytic enzyme systems of novel and efficient chitindegrading strains has become critical to achieving efficient bioconversion of chitin. In this study, multi-omics analysis revealed the chitinolytic hydrolytic and oxidative enzyme systems in Cellvibrio chitinivorans NN19, including four glycoside hydrolase family 18 chitinases Chi18 A/B/C/D, one glycoside hydrolase family 19 chitinase Chi19B, three glycoside hydrolase family 20 beta-N-acetylhexosaminidases Hex20 A/B/C, and four auxiliary activity family 10 lytic polysaccharide monooxygenases LPMO10 A/B/C. Meanwhile, gene knockout experiments targeting these chitinolytic enzymes demonstrated that the Chi18D-deficient strain completely lost its ability to degrade chitin, indicating that Chi18D is essential for initiating chitin depolymerization. Further enzymatic saccharification of chitin showed that the oxidative enzymes LPMO10A, LPMO10B, and LPMO10C significantly increased the yields of chitooligosaccharides from 0.69 mg/mL with Chi18D alone to 2.12 mg/mL, 1.51 mg/mL, and 2.70 mg/mL, respectively. Moreover, the combination of chitinolytic oxidative enzyme LPMO10C with Chi18D and Hex20A achieved a maximum chitooligosaccharides yield of 4.14 mg/mL for chitin degradation. These results demonstrated that the combination of LPMOs and beta-N-acetylhexosaminidases significantly enhance the degradation efficiency of chitinases, providing valuable insights for the development of efficient chitin-degrading enzyme systems.

Starch is a vital global commodity for food and industry, yet its agricultural production is unsustainable, relying on inefficient photosynthesis, arable land, and stable climates. Meanwhile, the annual production of ∼180 million tons of cellulose primarily from crop residues is largely wasted through incineration or landfilling. While enzymatic saccharification efficiently depolymerizes cellulose into glucose, its conversion to starch is fundamentally limited by the insufficient catalytic efficiency, poor substrate selectivity of key enzymes, and the reversibility of the core reactions they catalyze. Here, we optimized a cell-free three-enzyme cascade, designated the carbon-economical starch synthesis (CESS) pathway, to address this limitation. This in vitro cascade integrates three core enzymes: polyphosphate glucokinase (PPGK) for adenosine triphosphate (ATP)-free glucose activation, phosphoglucomutase (PGM), and α-glucan phosphorylase (αGP). We engineered the rate-limiting αGP from Thermotoga petrophila to generate the M7 variant (TpαGPM7), achieving a 4.54-fold higher catalytic efficiency, and optimized cofactors, identifying 20 mmol∙L−1 Mn2+ to critically shift PGM’s preference toward starch synthesis. The optimized pathway converts 5.4 g∙L−1 glucose (equivalent to 30 mmol∙L−1) to 3.50 g∙L−1 starch at 60 °C with a record-breaking 71.93% conversion yield and a space-time yield of 2.33 g∙L−1∙h−1, while minimizing carbon loss < 30%. The product was characterized as amylose. This photosynthesis-independent framework advances a sustainable route to valorize cellulosic waste into food-grade starch, offering a resilient strategy for food security and a circular bioeconomy.

The filamentous fungus Trichoderma reesei serves as an industrial workhorse for production of cellulolytic enzymes. However, the regulatory network governing cellulase biosynthesis in T. reesei remains incompletely understood, which limits rational engineering towards obtaining of hyper-producing strains. Herein, a previously uncharacterized NDT80/PhoG family transcription factor (TrNf1) was identified and characterized as a negative regulator of cellulase expression. Deletion of TrNf1 led to increases in the concentration of total secreted protein and overall cellulase activity by up to 4.2-fold and 3.1-fold, respectively. Deletion of TrNf1 induced global transcriptional reprogramming by activating a large array of carbohydrate-active enzyme (CAZyme) genes, remodeling the transcriptional regulatory network, and upregulating key activators such as xyr1 and ace3. Overexpressing xyr1 in the TrNf1-deleting strain substantially improved overall cellulase production by achieving a 3.2-fold of increase in the extracellular cellulase activity. Importantly, the repressive function is evolutionarily conserved in Aspergillus nidulans, where the homolog of TrNf1, i.e. AnNf1, similarly suppresses cellulase production. Collectively, our findings uncover a novel negative regulatory role for an NDT80/PhoG protein in fungal cellulase production and provide a conserved target for engineering the fungi into cellulase hyperproducers.

Microbial serine proteases are valuable for industrial applications due to broad substrate specificity and stability. However, heterologous overexpression in microbial hosts is often limited by cytotoxicity and poor secretion. This study developed an integrated strategy combining protein engineering and signal peptide optimization to enhance extracellular production of PrtA—a key acid-stable alkaline serine protease—in Komagataella phaffii. Directed evolution generated the Q245K variant, showing 1.35-fold higher extracellular expression than wild-type PrtA. A machine learning model, MPEPE (Mutation Predictor for Enhanced Protein Expression), was used to identify critical residues involved in protein secretion; saturation mutagenesis at the top-predicted site generated the I342D mutant with 1.48-fold improved productivity. The double mutant PrtA-Q245K/I342D achieved synergistic enhancement (1.84-fold higher secretion) without altering enzymatic properties. Evaluation of nine signal peptides revealed that serum albumin, α-factor (without pro-region), and PrtA’s native signal peptides each doubled the combinatorial mutant’s secretion, yielding 4.98-fold higher expression than the wild-type. In contrast, α-factor pro-region inclusion drastically reduced yields. In a 15-L fed-batch bioreactor, the optimized strain produced PrtA-Q245K/I342D at 4807.5 U/mL, equivalent to 1.5 g/L protein. The combined approach of directed evolution, machine learning-guided mutagenesis, and signal peptide engineering significantly boosted PrtA secretion while maintaining functional integrity. This strategy demonstrates strong potential for scalable industrial production of challenging heterologous proteases.

Since its inception, the CRISPR-Cas system, particularly Cas9, has demonstrated immense potential for life science applications, but expansion of the Cas9 toolkit is constrained by sequence-alignment-based strategies for mining and optimization. Here, we developed CasMiner-a deep-learning model for discovering and engineering novel Cas9 proteins. CasMiner achieved 99.63% accuracy in predicting Cas9s and identified VpCas9 from public databases. Experimental validation showed that VpCas9 exhibits robust double-strand cleavage activity. Combining CasMiner and evolutionary analysis, we engineered three mutants with markedly increased structural rigidity and positive charge. In vivo cleavage assays revealed that the mutant VPM2-3 achieved a higher average editing efficiency in rice callus and maize protoplasts than the wild-type VpCas9, the editing efficiency of which rivals that of SpCas9. This study thus establishes a comprehensive platform for mining and engineering Cas9 proteins, and provides VpCas9 and derivative nucleases as powerful tools that greatly broaden the horizon for genome-editing applications.

Electron transfer (ET) efficiency dictates catalytic performance in multi-domain self-sufficient cytochrome P450s. Conventional engineering, however, predominantly focuses on localized optimization of either the active-site pocket or linker regions, overlooking inter-domain conformational transitions and ET chain integrity. Herein, we report a holistic ET-optimization strategy integrating conformational dynamics modulation, ET pathway engineering, and substrate positioning tuning, which was applied to enhance ET in a chimeric P450 VK1-CYP116B46-L21 (L21) for calcifediol biosynthesis. [2Fe-2S]→Heme ET pathway engineering yielded variant L21-M2 (F346K/R354M), which decreased the conformational transition barrier by 4.5 kcal/mol and shortened the ET pathway by 4.05 Å, leading to a 72-fold enhancement in ET rate. Heme domain engineering generated variant L21-M3 (P83A/A177M/K180F), which shortened the near-attack conformations distance to 4.19 Å (from 4.62 Å) and increased reactive conformation to 38 % (from 25 %). The pentuple variant L21-M5 combined both improvements, which demonstrated exceptional catalytic performance: an 8.2- fold higher catalytic efficiency, a coupling efficiency of 56.78 %, and a total turnover number (TTN) of 3222. In a semi-preparative-scale biotransformation, L21-M5 achieved 3.26 g/L production of calcifediol with 82 % conversion, underscoring its strong industrial potential. These results highlight the efficacy of the proposed ET-optimization strategy and provide a transferable workflow for engineering multi-domain redox biocatalysts.

Limosilactobacillus reuteri Y15 (LRY15)-derived polyunsaturated fatty acids (PUFAs) restore follicular cell-cell junctions and fertility. LRY15, isolated from the intestine of alginate oligosaccharides (AOS)-dosed mice, improves ovarian function and fertility in cisplatin-induced subfertile mice. LRY15 produces PUFAs, whereas deletion of holo-acyl carrier protein synthase (AcpS) in LRY15 reduces PUFA synthesis and weakens its protective effects. Docosahexaenoic acid (DHA) supplementation recapitulates LRY15-mediated improvements in follicle and embryo development. Stereo-seq analysis reveals that LRY15 restores ovarian cell-cell junctions, suggesting junction remodeling as an important cellular process associated with LRY15 treatment. The Fads2-deficient mouse model further validates that LRY15-derived PUFAs restore follicular junction integrity and improve fertility.

As a facultative chemolithoautotrophic bacterium, Cupriavidus necator H16 uses the Entner-Doudoroff (ED) pathway for heterotrophic growth on carbohydrates such as fructose and the Calvin cycle for lithoautotrophic carbon dioxide fixation. In a previous study, we found that an ED pathway-deficient C. necator strain can survive on fructose, but the underlying metabolic pathway remained unclear. This study aimed to elucidate the metabolic mechanism of fructose metabolism in this ED pathway-deficient C. necator strain. First, the metabolic characteristics of fructose catabolism in the deficient strain were examined. Then, the roles of glycolysis/gluconeogenesis, the Calvin shunt, and the non-oxidative pentose phosphate pathway (non-OxPPP) in the metabolism of fructose were identified through comparative transcriptomic analysis combined with 13C tracer experiments. Further growth experiments using knockout strains of key genes involved in these pathways confirmed that the non-OxPPP compensates for the blocked ED pathway to metabolize fructose and provide a precursor for the Calvin shunt, thereby driving subsequent carbon fluxes. Additionally, phosphoglycolate salvage pathways, particularly the malate cycle, are crucial for recycling glycolate-2-phosphate produced during RuBisCO-catalyzed oxidation. This study revealed a novel fructose metabolism mechanism in C. necator and highlighted its metabolic flexibility, thereby deepening our understanding of its carbon utilization strategies and providing a theoretical basis for further metabolic engineering research.

Tartaric acid (TA) is a crucial hydroxy-carboxylic acid chelating agent extensively employed in the pharmaceutical, food, dye, and energy industries. The utilization of glucose oxidation for TA synthesis represents an appealing yet challenging, environmentally-friendly approach. In this study, we developed a two-step chemoenzymatic strategy involving the efficient generation of gluconic acid (GA) through glucose oxidase (Gox) catalysis, followed by chemical decarboxylation to synthesize TA from GA. To optimize the production efficiency of GA and eliminate substrate inhibition in chemocatalysis, we successfully engineered a quadruple variant GoxM10 by implementing rigidifying structure strategies and refining the pH curve. This resulted in a substantial increase in the conversion rate of GA from 45 ​% to 100 ​%, along with improved thermo- and pH stability, as well as enhanced catalytic efficiency. Subsequently, durable bimetallic catalysts were designed for the conversion of Gox-bioconverted GA (bio-GA) to TA. Given the presence of impurities such as acetic acid and sodium acetate in bio-GA, which impacted the selectivity of TA, our focus was on investigating the influence of impurities in bio-GA on the performance of a bimetallic AuPt/TiO2 catalyst and elucidating an inhibition mechanism that reveals a strong interaction between acetate and Au site, resulting in the formation of a complex compound while promoting uncontrollable C-C cleavage reactions. Through employing a chemoenzymatic two-step strategy, we successfully achieved an environmentally friendly and sustainable route from glucose to TA. This study provides valuable insights into catalyst design and optimization of reaction routes for renewable carboxylic acids and their derivatives synthesized through chemoenzymatic methods.

Bioactive peptides, defined as amino acid chains exhibiting diverse biological functions such as antimicrobial, antioxidant, and anti-inflammatory activities, are primarily generated through protein digestion methods including enzymatic hydrolysis, physical processing techniques and controlled microbial fermentation. Conventional discovery techniques that rely on multi-stage separation processes, such as enzymatic digestion, ultrafiltration, ion-exchange chromatography, gel filtration chromatography, and reverse-phase high-performance liquid chromatography (RP-HPLC) inherently demand substantial laboratory resources and extended timeframes. To address these limitations, artificial intelligence (AI)-driven approaches have emerged as transformative discovery platforms. These computational pipelines systematically execute six critical phases: comprehensive data acquisition and curation, advanced feature engineering utilizing physicochemical descriptors, machine learning model construction using algorithms, iterative model training incorporating hyperparameter optimization, rigorous validation against benchmark datasets, and high-throughput bioactive peptide prediction. This comprehensive review critically evaluates recent AI applications across four key bioactive peptide categories including antimicrobial peptides, antioxidant peptides, anti-inflammatory peptides, and multifunctional variants. Furthermore, it proposes integrated enhancement strategies such as classifying peptides via their functional mechanism or using database-independent modeling approaches. Additionally, based on AI methods, scenario-specific peptide customization and prediction of bioactivity in digested proteomes are anticipated to be achieved in the future.