Chengxi Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Chengxi Li has a total of 91 co-authors. Here are the top co-authors by number of shared publications:
Chengrui Hu
8 shared publicationsXingxing Yang
Key Laboratory of Biomass Chemical Engineering of Ministry of Education, College of Chemical and Biological Engineering
8 shared publicationsYourong Pan
Key Laboratory of Biomass Chemical Engineering of Ministry of Education, College of Chemical and Biological Engineering, Zhejiang University, Hangzhou 310027, China
5 shared publications- Bradley Lether Pentelute
Department of Chemistry Massachusetts Institute of Technology Cambridge Massachusetts USA
4 shared publications Cheng Ren
Zhejiang University
4 shared publicationsFeng Wan
Zhejiang University
4 shared publications
Latest Publications
- 15 November 2026
- Published by Wiley in Angewandte Chemie
A Phosphonium Relay for Integrated Cleavage‐to‐Cyclization of Cyclic Peptides
Head‐to‐tail macrocyclization of peptides is typically performed through stepwise workflows that separate resin cleavage, isolation of linear intermediates, and subsequent cyclization, largely due to incompatible activation requirements for each transformation. Here we report a phosphonium‐mediated relay that unifies these steps within a single activation node. In this design, an electrophile‐activated phosphonium species promotes cleavage from standard 2‐chlorotrityl chloride (2‐CTC) resin and is proposed to enable in situ formation of a C ‐terminally activated intermediate for intramolecular amidation. The strategy directly couples cleavage to cyclization on standard peptide supports without auxiliary linkers, intermediate purification, or high‐dilution cyclization conditions. The method is compatible with diverse peptide sequccnces and functional handles and can be integrated into a closed‐loop automated workflow for end‐to‐end synthesis, with advantages for hydrophobic substrates by avoiding isolation of poorly soluble linear intermediates such as Vespid chemotactic peptide 5f . Process‐level benchmarking against side‐chain anchoring further demonstrated favorable process metrics and gram‐scale scalability. This work establishes a linker‐free cleavage‐to‐cyclization strategy and provides a practical framework for integrated and automation‐compatible cyclic peptide synthesis.
- 15 November 2026
- Published by Wiley in Angewandte Chemie
A Phosphonium Relay for Integrated Cleavage‐to‐Cyclization of Cyclic Peptides
Head‐to‐tail macrocyclization of peptides is typically performed through stepwise workflows that separate resin cleavage, isolation of linear intermediates, and subsequent cyclization, largely due to incompatible activation requirements for each transformation. Here we report a phosphonium‐mediated relay that unifies these steps within a single activation node. In this design, an electrophile‐activated phosphonium species promotes cleavage from standard 2‐chlorotrityl chloride (2‐CTC) resin and is proposed to enable in situ formation of a C ‐terminally activated intermediate for intramolecular amidation. The strategy directly couples cleavage to cyclization on standard peptide supports without auxiliary linkers, intermediate purification, or high‐dilution cyclization conditions. The method is compatible with diverse peptide sequccnces and functional handles and can be integrated into a closed‐loop automated workflow for end‐to‐end synthesis, with advantages for hydrophobic substrates by avoiding isolation of poorly soluble linear intermediates such as Vespid chemotactic peptide 5f . Process‐level benchmarking against side‐chain anchoring further demonstrated favorable process metrics and gram‐scale scalability. This work establishes a linker‐free cleavage‐to‐cyclization strategy and provides a practical framework for integrated and automation‐compatible cyclic peptide synthesis.
- 15 November 2026
- Published by Wiley in Angewandte Chemie International Edition
A Phosphonium Relay for Integrated Cleavage‐to‐Cyclization of Cyclic Peptides
Head‐to‐tail macrocyclization of peptides is typically performed through stepwise workflows that separate resin cleavage, isolation of linear intermediates, and subsequent cyclization, largely due to incompatible activation requirements for each transformation. Here we report a phosphonium‐mediated relay that unifies these steps within a single activation node. In this design, an electrophile‐activated phosphonium species promotes cleavage from standard 2‐chlorotrityl chloride (2‐CTC) resin and is proposed to enable in situ formation of a C ‐terminally activated intermediate for intramolecular amidation. The strategy directly couples cleavage to cyclization on standard peptide supports without auxiliary linkers, intermediate purification, or high‐dilution cyclization conditions. The method is compatible with diverse peptide sequccnces and functional handles and can be integrated into a closed‐loop automated workflow for end‐to‐end synthesis, with advantages for hydrophobic substrates by avoiding isolation of poorly soluble linear intermediates such as Vespid chemotactic peptide 5f . Process‐level benchmarking against side‐chain anchoring further demonstrated favorable process metrics and gram‐scale scalability. This work establishes a linker‐free cleavage‐to‐cyclization strategy and provides a practical framework for integrated and automation‐compatible cyclic peptide synthesis.
- 15 November 2026
- Published by Wiley in Angewandte Chemie International Edition
A Phosphonium Relay for Integrated Cleavage‐to‐Cyclization of Cyclic Peptides
Head‐to‐tail macrocyclization of peptides is typically performed through stepwise workflows that separate resin cleavage, isolation of linear intermediates, and subsequent cyclization, largely due to incompatible activation requirements for each transformation. Here we report a phosphonium‐mediated relay that unifies these steps within a single activation node. In this design, an electrophile‐activated phosphonium species promotes cleavage from standard 2‐chlorotrityl chloride (2‐CTC) resin and is proposed to enable in situ formation of a C ‐terminally activated intermediate for intramolecular amidation. The strategy directly couples cleavage to cyclization on standard peptide supports without auxiliary linkers, intermediate purification, or high‐dilution cyclization conditions. The method is compatible with diverse peptide sequccnces and functional handles and can be integrated into a closed‐loop automated workflow for end‐to‐end synthesis, with advantages for hydrophobic substrates by avoiding isolation of poorly soluble linear intermediates such as Vespid chemotactic peptide 5f . Process‐level benchmarking against side‐chain anchoring further demonstrated favorable process metrics and gram‐scale scalability. This work establishes a linker‐free cleavage‐to‐cyclization strategy and provides a practical framework for integrated and automation‐compatible cyclic peptide synthesis.
- Open Access
- 30 September 2026
- Published by American Chemical Society (ACS) in ChemRxiv
Linker-Encoded Enzymatic Topology Control Enables Automated Peptide Stapling with Reusable Immobilized mTGase
Microbial transglutaminase (mTGase) readily forms intermolecular isopeptide bonds, but this native bias promotes peptide oligomerization at synthetically useful concentrations. Here we encode reaction topology in a minimal diamine linker to redirect mTGase toward intramolecular stapling of unprotected glutamine-containing peptides. Direct comparison with a linker-free Lys–Gln substrate shows that the linker-encoded architecture maintains macrocyclization as the dominant pathway from 0.25 to 2.0 mM, whereas the native substrate undergoes exclusively intermolecular ligation. Linker and peptide studies establish geometric, electronic, and sequence requirements, and molecular simulations are consistent with the organization of the acyl–enzyme intermediate. The resulting staples serve as versatile synthetic handles for orthogonal bicyclization, late-stage functionalization, and reductive ring opening. The covalent immobilization of mTGase enables catalyst recycling and seamless integration into a continuous, automated peptide-stapling platform capable of processing diverse substrates. This linker-encoded strategy provides a route to control enzymatic reaction topology for programmable constrained-peptide synthesis.
- 30 September 2026
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessLinker-Encoded Enzymatic Topology Control Enables Automated Peptide Stapling with Reusable Immobilized mTGase
Microbial transglutaminase (mTGase) readily forms intermolecular isopeptide bonds, but this native bias promotes peptide oligomerization at synthetically useful concentrations. Here we encode reaction topology in a minimal diamine linker to redirect mTGase toward intramolecular stapling of unprotected glutamine-containing peptides. Direct comparison with a linker-free Lys–Gln substrate shows that the linker-encoded architecture maintains macrocyclization as the dominant pathway from 0.25 to 2.0 mM, whereas the native substrate undergoes exclusively intermolecular ligation. Linker and peptide studies establish geometric, electronic, and sequence requirements, and molecular simulations are consistent with the organization of the acyl–enzyme intermediate. The resulting staples serve as versatile synthetic handles for orthogonal bicyclization, late-stage functionalization, and reductive ring opening. The covalent immobilization of mTGase enables catalyst recycling and seamless integration into a continuous, automated peptide-stapling platform capable of processing diverse substrates. This linker-encoded strategy provides a route to control enzymatic reaction topology for programmable constrained-peptide synthesis.
- Open Access
- 8 September 2026
- Published by American Chemical Society (ACS) in JACS Au
- Vol. 6 (9) , 5264-5274
Secondary-Amine Relay Enables Lysine-Directed Multicomponent Functionalization of Peptides
Multicomponent reactions provide powerful modularity for peptide functionalization, but their application to lysine residues remains limited by the low efficiency of direct primary-amine-based activation pathways under peptide-compatible conditions. Here we report a secondary-amine relay strategy for lysine-directed multicomponent functionalization of peptides. In this approach, salicylaldehyde derivatives, arylboronic acids, and a cyclic secondary amine are combined to generate a transient electrophilic intermediate, which is subsequently intercepted by the ε-amine of lysine-containing peptides. By separating electrophile generation from direct lysine activation, this relay protocol improves the efficiency of lysine modification while retaining the modularity of multicomponent chemistry. Mechanistic experiments, isolated intermediate studies, kinetic comparisons, and density functional theory calculations support a pathway involving secondary-amine-mediated intermediate formation followed by lysine capture. The method enables modification of diverse bioactive peptides, incorporation of drug-derived and fluorescent aldehyde or boronic acid components, and lysine–lysine stapling using bifunctional aldehydes. The reaction is further compatible with automated and parallel solid-phase peptide modification workflows, providing rapid access to libraries of lysine-functionalized peptide conjugates. This study provides a practical secondary-amine relay platform for modular peptide functionalization and expands the utility of multicomponent chemistry in automated peptide synthesis.
- 8 September 2026
- Published by American Chemical Society (ACS) in JACS Au
- Vol. 6 (9) , 5264-5274
Open AccessSecondary-Amine Relay Enables Lysine-Directed Multicomponent Functionalization of Peptides
Multicomponent reactions provide powerful modularity for peptide functionalization, but their application to lysine residues remains limited by the low efficiency of direct primary-amine-based activation pathways under peptide-compatible conditions. Here we report a secondary-amine relay strategy for lysine-directed multicomponent functionalization of peptides. In this approach, salicylaldehyde derivatives, arylboronic acids, and a cyclic secondary amine are combined to generate a transient electrophilic intermediate, which is subsequently intercepted by the ε-amine of lysine-containing peptides. By separating electrophile generation from direct lysine activation, this relay protocol improves the efficiency of lysine modification while retaining the modularity of multicomponent chemistry. Mechanistic experiments, isolated intermediate studies, kinetic comparisons, and density functional theory calculations support a pathway involving secondary-amine-mediated intermediate formation followed by lysine capture. The method enables modification of diverse bioactive peptides, incorporation of drug-derived and fluorescent aldehyde or boronic acid components, and lysine–lysine stapling using bifunctional aldehydes. The reaction is further compatible with automated and parallel solid-phase peptide modification workflows, providing rapid access to libraries of lysine-functionalized peptide conjugates. This study provides a practical secondary-amine relay platform for modular peptide functionalization and expands the utility of multicomponent chemistry in automated peptide synthesis.
- 10 August 2026
- Published by American Chemical Society (ACS) in Journal of the American Chemical Society
SuFEx-Enabled Acyl Fluoride Chemistry for High-Concentration Minimal-Oligomerization Peptide Macrocyclization
Head-to-tail macrocyclization is a cornerstone transformation in cyclic peptide synthesis, yet its practical implementation remains constrained by extensive oligomerization under concentrated conditions. Here we report a SuFEx-enabled acyl fluoride activation strategy that intrinsically biases peptide macrocyclization toward intramolecular amide bond formation, enabling highly selective head-to-tail cyclization at concentrations up to 100 mM. In contrast to conventional uronium- and phosphonium-based coupling reagents, which generate complex oligomeric mixtures under identical conditions, the SuFEx-derived acyl fluoride pathway consistently affords cyclic monomers with minimal dimerization across diverse peptide sequences. Density functional theory calculations reveal that acyl fluoride activation reshapes the macrocyclization energy landscape by favoring irreversible ring closure while disfavoring oligomer growth, providing a mechanistic basis for the observed chemoselectivity. The method exhibits broad substrate scope, tolerating variations in ring size, sequence composition, and functionalized residues, and is readily translated to solid-phase and automated synthesis platforms. These results establish SuFEx-enabled acyl fluoride activation as a fundamentally distinct and practically robust solution to peptide macrocyclization under high-concentration conditions.
- 10 August 2026
- Published by American Chemical Society (ACS) in Journal of the American Chemical Society
SuFEx-Enabled Acyl Fluoride Chemistry for High-Concentration Minimal-Oligomerization Peptide Macrocyclization
Head-to-tail macrocyclization is a cornerstone transformation in cyclic peptide synthesis, yet its practical implementation remains constrained by extensive oligomerization under concentrated conditions. Here we report a SuFEx-enabled acyl fluoride activation strategy that intrinsically biases peptide macrocyclization toward intramolecular amide bond formation, enabling highly selective head-to-tail cyclization at concentrations up to 100 mM. In contrast to conventional uronium- and phosphonium-based coupling reagents, which generate complex oligomeric mixtures under identical conditions, the SuFEx-derived acyl fluoride pathway consistently affords cyclic monomers with minimal dimerization across diverse peptide sequences. Density functional theory calculations reveal that acyl fluoride activation reshapes the macrocyclization energy landscape by favoring irreversible ring closure while disfavoring oligomer growth, providing a mechanistic basis for the observed chemoselectivity. The method exhibits broad substrate scope, tolerating variations in ring size, sequence composition, and functionalized residues, and is readily translated to solid-phase and automated synthesis platforms. These results establish SuFEx-enabled acyl fluoride activation as a fundamentally distinct and practically robust solution to peptide macrocyclization under high-concentration conditions.
- 8 July 2026
- Published by American Chemical Society (ACS) in Journal of the American Chemical Society
- Vol. 148 (28) , 30476-30485
A Modular PMDA Linker Enables Lysine-Selective Cyclization of Unprotected Peptides and Automated Macrocycle Assembly
Macrocyclic peptide stapling is a powerful strategy to enhance conformational stability and functional performance, yet existing approaches often suffer from limited chemoselectivity, constrained modularity, or poor compatibility with automated synthesis. Here we report a pyromellitic dianhydride (PMDA)-based Lys-Lys stapling platform that enables highly chemoselective intramolecular bis-amidation of fully unprotected peptides under mild conditions. Distinct from conventional bifunctional linkers, PMDA operates as a tetrafunctional scaffold, simultaneously inducing efficient macrocyclization while installing two carboxylate handles for downstream diversification. The reaction exhibits exceptional selectivity for lysine ε-amines, tolerating all other nucleophilic side chains, including cysteine, across a broad range of peptide sequences and linker spacings. The resulting macrocycles support stepwise bicyclization and modular postcyclization conjugation with nucleic acids, small-molecule drugs, and affinity tags. Application to antimicrobial peptides demonstrates enhanced conformational constraint and markedly improved proteolytic stability. Importantly, the chemistry is intrinsically compatible with solid-phase peptide synthesis and fully integrated, end-to-end automated workflows without hardware modification. This PMDA-mediated stapling strategy establishes a versatile, automation-ready linker platform for rapid assembly and functional diversification of macrocyclic peptides.
- 8 July 2026
- Published by American Chemical Society (ACS) in Journal of the American Chemical Society
- Vol. 148 (28) , 30476-30485
A Modular PMDA Linker Enables Lysine-Selective Cyclization of Unprotected Peptides and Automated Macrocycle Assembly
Macrocyclic peptide stapling is a powerful strategy to enhance conformational stability and functional performance, yet existing approaches often suffer from limited chemoselectivity, constrained modularity, or poor compatibility with automated synthesis. Here we report a pyromellitic dianhydride (PMDA)-based Lys-Lys stapling platform that enables highly chemoselective intramolecular bis-amidation of fully unprotected peptides under mild conditions. Distinct from conventional bifunctional linkers, PMDA operates as a tetrafunctional scaffold, simultaneously inducing efficient macrocyclization while installing two carboxylate handles for downstream diversification. The reaction exhibits exceptional selectivity for lysine ε-amines, tolerating all other nucleophilic side chains, including cysteine, across a broad range of peptide sequences and linker spacings. The resulting macrocycles support stepwise bicyclization and modular postcyclization conjugation with nucleic acids, small-molecule drugs, and affinity tags. Application to antimicrobial peptides demonstrates enhanced conformational constraint and markedly improved proteolytic stability. Importantly, the chemistry is intrinsically compatible with solid-phase peptide synthesis and fully integrated, end-to-end automated workflows without hardware modification. This PMDA-mediated stapling strategy establishes a versatile, automation-ready linker platform for rapid assembly and functional diversification of macrocyclic peptides.
- 18 June 2026
- Published by Elsevier in Trends in Biochemical Sciences
- Vol. 51 (8) , 844-845
CycloPepper learns cyclization sites in therapeutic peptides
- 18 June 2026
- Published by Elsevier in Trends in Biochemical Sciences
- Vol. 51 (8) , 844-845
CycloPepper learns cyclization sites in therapeutic peptides
- 13 May 2026
- Published by Wiley in Angewandte Chemie
- Vol. 138 (27)
SPECTRAL: An Intelligent and Ultra‐Sensitive Photonic Hydrogel Platform for Biomarker‐Based Cancer Prediction
Point‐of‐care (POC) biosensing holds great promise for noninvasive cancer diagnostics, as circulating tumor DNA (ctDNA) serves as a minimally invasive biomarker for early detection. However, effective ctDNA detection remains challenging due to low abundance and complex workflows. Herein, we present SPECTRAL (Smart Photonic‐hydrogel Enhanced chip for Cancer Type Recognition and AnaLysis), a sequencing‐free platform integrating photonic‐crystal hydrogels with charge‐neutral peptide nucleic acid (PNA) probes for efficient volumetric signal transduction and single‐nucleotide discrimination. Photonic‐crystal‐enhanced fluorescence amplifies signals of fluorescently labeled recombinase polymerase amplification (RPA) amplicons, achieving a detection limit of 100 copies per µL, with substantially improved sensitivity over conventional fluorescence assays. A single SPECTRAL chip simultaneously profiles 205 ctDNA mutations and eight protein biomarkers from plasma samples. The PNA sensor library is generated via an automated synthesis system guided by machine learning (ML)‐predicted probe difficulty. Multimodal signals are analyzed using cloud‐based ML to classify multiple cancer types (lung, breast, and colorectal), achieving 90.0% specificity and 86.7% sensitivity (87.5% overall accuracy) within 100 min from sample to diagnosis, based on validation with patient plasma samples. By combining molecular biosensing, photonic signal amplification, and AI‐driven analysis, SPECTRAL provides a low‐instrument‐dependent platform for decentralized liquid‐biopsy cancer diagnostics.
- 13 May 2026
- Published by Wiley in Angewandte Chemie
- Vol. 138 (27)
SPECTRAL: An Intelligent and Ultra‐Sensitive Photonic Hydrogel Platform for Biomarker‐Based Cancer Prediction
Point‐of‐care (POC) biosensing holds great promise for noninvasive cancer diagnostics, as circulating tumor DNA (ctDNA) serves as a minimally invasive biomarker for early detection. However, effective ctDNA detection remains challenging due to low abundance and complex workflows. Herein, we present SPECTRAL (Smart Photonic‐hydrogel Enhanced chip for Cancer Type Recognition and AnaLysis), a sequencing‐free platform integrating photonic‐crystal hydrogels with charge‐neutral peptide nucleic acid (PNA) probes for efficient volumetric signal transduction and single‐nucleotide discrimination. Photonic‐crystal‐enhanced fluorescence amplifies signals of fluorescently labeled recombinase polymerase amplification (RPA) amplicons, achieving a detection limit of 100 copies per µL, with substantially improved sensitivity over conventional fluorescence assays. A single SPECTRAL chip simultaneously profiles 205 ctDNA mutations and eight protein biomarkers from plasma samples. The PNA sensor library is generated via an automated synthesis system guided by machine learning (ML)‐predicted probe difficulty. Multimodal signals are analyzed using cloud‐based ML to classify multiple cancer types (lung, breast, and colorectal), achieving 90.0% specificity and 86.7% sensitivity (87.5% overall accuracy) within 100 min from sample to diagnosis, based on validation with patient plasma samples. By combining molecular biosensing, photonic signal amplification, and AI‐driven analysis, SPECTRAL provides a low‐instrument‐dependent platform for decentralized liquid‐biopsy cancer diagnostics.
- 11 May 2026
- Published by Wiley in Angewandte Chemie International Edition
- Vol. 65 (27) , e7454058
SPECTRAL: An Intelligent and Ultra‐Sensitive Photonic Hydrogel Platform for Biomarker‐Based Cancer Prediction
Point‐of‐care (POC) biosensing holds great promise for noninvasive cancer diagnostics, as circulating tumor DNA (ctDNA) serves as a minimally invasive biomarker for early detection. However, effective ctDNA detection remains challenging due to low abundance and complex workflows. Herein, we present SPECTRAL (Smart Photonic‐hydrogel Enhanced chip for Cancer Type Recognition and AnaLysis), a sequencing‐free platform integrating photonic‐crystal hydrogels with charge‐neutral peptide nucleic acid (PNA) probes for efficient volumetric signal transduction and single‐nucleotide discrimination. Photonic‐crystal‐enhanced fluorescence amplifies signals of fluorescently labeled recombinase polymerase amplification (RPA) amplicons, achieving a detection limit of 100 copies per µL, with substantially improved sensitivity over conventional fluorescence assays. A single SPECTRAL chip simultaneously profiles 205 ctDNA mutations and eight protein biomarkers from plasma samples. The PNA sensor library is generated via an automated synthesis system guided by machine learning (ML)‐predicted probe difficulty. Multimodal signals are analyzed using cloud‐based ML to classify multiple cancer types (lung, breast, and colorectal), achieving 90.0% specificity and 86.7% sensitivity (87.5% overall accuracy) within 100 min from sample to diagnosis, based on validation with patient plasma samples. By combining molecular biosensing, photonic signal amplification, and AI‐driven analysis, SPECTRAL provides a low‐instrument‐dependent platform for decentralized liquid‐biopsy cancer diagnostics.
- 11 May 2026
- Published by Wiley in Angewandte Chemie International Edition
- Vol. 65 (27) , e7454058
SPECTRAL: An Intelligent and Ultra‐Sensitive Photonic Hydrogel Platform for Biomarker‐Based Cancer Prediction
Point‐of‐care (POC) biosensing holds great promise for noninvasive cancer diagnostics, as circulating tumor DNA (ctDNA) serves as a minimally invasive biomarker for early detection. However, effective ctDNA detection remains challenging due to low abundance and complex workflows. Herein, we present SPECTRAL (Smart Photonic‐hydrogel Enhanced chip for Cancer Type Recognition and AnaLysis), a sequencing‐free platform integrating photonic‐crystal hydrogels with charge‐neutral peptide nucleic acid (PNA) probes for efficient volumetric signal transduction and single‐nucleotide discrimination. Photonic‐crystal‐enhanced fluorescence amplifies signals of fluorescently labeled recombinase polymerase amplification (RPA) amplicons, achieving a detection limit of 100 copies per µL, with substantially improved sensitivity over conventional fluorescence assays. A single SPECTRAL chip simultaneously profiles 205 ctDNA mutations and eight protein biomarkers from plasma samples. The PNA sensor library is generated via an automated synthesis system guided by machine learning (ML)‐predicted probe difficulty. Multimodal signals are analyzed using cloud‐based ML to classify multiple cancer types (lung, breast, and colorectal), achieving 90.0% specificity and 86.7% sensitivity (87.5% overall accuracy) within 100 min from sample to diagnosis, based on validation with patient plasma samples. By combining molecular biosensing, photonic signal amplification, and AI‐driven analysis, SPECTRAL provides a low‐instrument‐dependent platform for decentralized liquid‐biopsy cancer diagnostics.
- Open Access
- 14 February 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
CycloPepper: a machine learning platform for predicting cyclization outcomes and optimizing synthesis of therapeutic cyclopeptides
Cyclic peptides exhibit remarkable stability, membrane permeability, and binding affinity, positioning them as promising therapeutics. However, their synthesis, particularly on-resin head-to-tail cyclization, remains challenging, with cyclization site selection critically influencing yield. Here, we introduce a machine learning (ML) approach to predict cyclization outcomes, leveraging CycloBot, our fully automated cyclic peptide synthesis platform. Using this system, we generate a standardized dataset of 306 cyclic peptides (2–14 residues) and develop an ML model achieving an average prediction accuracy of 84%. Experimental validation with 74 random and therapeutic peptides showed an 86% prediction consistency. To facilitate practical use, we built CycloPepper, a user-friendly platform available through both web and software interfaces, enabling rapid cyclization site assessment. This tool effectively identified potential cyclization sites for disease-targeting peptides, including cancer biomarkers. Our work illustrates the potential of ML-assisted synthesis to streamline cyclic peptide synthesis and accelerate therapeutic discovery.
- 14 February 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Open AccessCycloPepper: a machine learning platform for predicting cyclization outcomes and optimizing synthesis of therapeutic cyclopeptides
Cyclic peptides exhibit remarkable stability, membrane permeability, and binding affinity, positioning them as promising therapeutics. However, their synthesis, particularly on-resin head-to-tail cyclization, remains challenging, with cyclization site selection critically influencing yield. Here, we introduce a machine learning (ML) approach to predict cyclization outcomes, leveraging CycloBot, our fully automated cyclic peptide synthesis platform. Using this system, we generate a standardized dataset of 306 cyclic peptides (2–14 residues) and develop an ML model achieving an average prediction accuracy of 84%. Experimental validation with 74 random and therapeutic peptides showed an 86% prediction consistency. To facilitate practical use, we built CycloPepper, a user-friendly platform available through both web and software interfaces, enabling rapid cyclization site assessment. This tool effectively identified potential cyclization sites for disease-targeting peptides, including cancer biomarkers. Our work illustrates the potential of ML-assisted synthesis to streamline cyclic peptide synthesis and accelerate therapeutic discovery.