CycPeptAI · Computational Design & Experimental Planning

Cyclization & Synthetic Feasibility

Design viable cyclization sites and bicyclic topologies within a defined linkage chemistry—and carry the proposed precursor, reaction order, failure risks and topology-confirmation plan into the laboratory.

Discuss Your Cyclization Project

Overview

A Cyclic Peptide Is More Than a Circular Sequence

A proposed ring is useful only when its connectivity can be made, isolated and verified. The same nominal cyclic peptide can arise from different linear precursors, closure positions, protecting-group schemes and reaction orders. These choices change steric accessibility, side reactions, product mixtures and the evidence needed to assign the final structure.

Our service starts by fixing the chemical design space with the client. We specify the allowed linkage type, reactive residues, linker class, orthogonal protection strategy and assay-relevant constraints before computational ranking. Within that scope, we compare candidate ring-closure positions or bicyclic architectures, propose a precursor and stepwise route, flag chemical liabilities and define analytical tests for the intended topology. The outcome is a decision package for a focused synthesis pilot, not a promise that every proposed ring will close successfully.

This is particularly useful when a binding-oriented design campaign yields several attractive peptide sequences but no clear route to the corresponding macrocycles. It also applies when a known linear or monocyclic hit needs an additional constraint without destroying the binding epitope. The work can connect to AI-accelerated library design and lead optimization, while retaining a distinct focus on reaction feasibility and structural identity.

Scope first: a prediction trained on head-to-tail amide closure under one on-resin protocol should not be treated as a general success score for thiol alkylation, click chemistry, lactamization or a different linker. We document the supported chemistry and mark transfers outside that domain as hypotheses for experimental testing.

Core Services

Four Decisions Before a Synthesis Campaign

01

Linkage-Constrained Cyclization Site Design

Input: peptide sequence or candidate library, permitted amide or side-chain linkage, reactive handles, stereochemistry and known binding residues. We enumerate chemically valid closure pairs and linear disconnections within that fixed reaction class. Conformational sampling, ring-size checks, local steric analysis and known motif liabilities support a comparative ranking. The output is a site-by-site rationale with excluded positions and uncertainties, rather than an unrestricted list of theoretical rings.

For head-to-tail chemistry, changing the linearization point can change the cyclization outcome even though the final cyclic sequence is nominally identical. We therefore describe the actual linear precursor and closure bond for every proposed candidate. When the peptide contains a sensitive epitope or a pharmacophore, we prioritize closing away from residues whose modification may impair activity.

02

Bicyclic Topology & Linker Geometry

Input: allowed anchor residues, linker valency, intended loop lengths and required orientation of binding motifs. We compare feasible graph connectivities, attachment orders and linker geometries within the approved chemistry. A three-anchor scaffold and a two-bridge sequence are distinct reaction problems: each needs its own mapping of anchor-to-arm correspondence and potential regioisomers.

The design output identifies the intended topology explicitly, including which residues are joined by each bridge, how many rings result and which alternative connectivities may be produced. Structural models can help reject severe strain and prioritize candidates, but the model alone cannot establish the product connectivity or the distribution of topological isomers after synthesis.

03

Precursor, Protection & Connection Order

Input: available monomers, reactive groups, linker inventory, resin or solution-phase preferences, cleavage constraints and production scale. We define a proposed linear precursor, protection and deprotection sequence, coupling or crosslinking steps, and the order of ring closures. Where orthogonal transformations are contemplated, we state which handle is exposed at each step and which side reactions are anticipated.

We compare one-pot and stepwise routes only when the selected reaction chemistry allows a meaningful comparison. For a bicyclic candidate, the route notes whether the first ring may constrain access to the second closure or whether premature second reaction could yield the wrong connectivity. Final reagent conditions are set during experimental development, not inferred from a sequence-level design score.

04

Failure-Risk & Verification Plan

Input: candidate structures, proposed route and any pilot LC–MS, chromatographic or yield data. We score risks such as poor intramolecular closure, oligomerization, epimerization, side-chain competition, protecting-group incompatibility, linker overreaction, insolubility and ambiguous isomers. Each risk is coupled to a mitigation experiment or a decision to deprioritize the design.

The output includes a topology-confirmation matrix. Intact mass and purity are necessary early checks but are often insufficient to distinguish constitutional or regioisomers. Depending on chemistry, orthogonal cleavage, targeted fragmentation, MS/MS mapping, NMR, comparison with independently prepared standards and analytical separation can help establish the intended connections. The exact package is agreed for the question and available material.

Project Inputs & Boundaries

What We Need to Define a Testable Design Space

Sequence and Biological Context

Provide the starting linear, cyclic or bicyclic sequence; modifications and stereochemical assignments; the target and known or hypothesized binding residues; and any affinity, activity or selectivity data. If binding information is uncertain, we label the proposed protected positions as hypotheses. An existing structure or conformer ensemble is helpful but not required for an initial route review.

Permitted Chemistry

Specify whether the campaign uses head-to-tail amide closure, selected side-chain lactamization, cysteine–linker crosslinking, a defined click pair or another validated reaction. Share allowed linkers, reactive residues, protected monomers, cleavage conditions, synthesis platform and excluded modifications. If the chemistry is not yet chosen, an initial feasibility review compares a small number of plausible classes before any class-specific scoring.

Experimental Evidence

Share crude and purified LC–MS traces, isolated yields, major side products, reaction logs, analytical standards and topology assignments where available. A few consistent examples under the same chemistry are more useful for local calibration than a larger heterogeneous set that combines incompatible definitions of success. We record missing values and changing protocols before training or updating any predictive model.

The service is scoped to specified peptide and linkage chemistry. Biological potency, membrane permeability and oral exposure are separate endpoints. Cyclization can alter them, but a chemically viable ring does not establish a therapeutically useful profile. Those endpoints can be addressed through the site's permeability and ADMET services when relevant.

Integrated Workflow

From Defined Chemistry to Confirmed Topology

Each stage yields a document or experimental gate. A client may commission the whole sequence or begin with a route audit for an existing hit.

StageKey ActivitiesDecision Output
1. Scope chemistryFreeze linkage type, handles, linker inventory, peptide format, assay priorities and acceptance criteria. Separate candidate potency from reaction success.Written design constraints and an applicability statement for computational methods.
2. Enumerate closure sitesGenerate valid precursors and closure pairs; inspect ring size, local conformation, epimerization-sensitive residues and steric exposure.Ranked site–precursor map with excluded disconnections and stated uncertainty.
3. Map bicyclic topologyAssign anchor-to-linker connections, loop lengths, possible regioisomers and a permissible first-to-second closure sequence.Topology diagrams and route alternatives tied to a precise molecular identity.
4. Review route risksCheck protection, deprotection, coupling order, competing nucleophiles, intramolecular versus intermolecular reaction and purification burden.Risk register, mitigation experiments and a small pilot synthesis set.
5. Pilot and characterizeWhere included in scope, synthesize a limited set and evaluate conversion, impurity profile, purity, and orthogonal structural evidence.Observed feasibility and a confidence-qualified topology assignment, or a documented unresolved ambiguity.
6. Learn and redesignCompare predictions with measured outcomes, review failures and refine only within the relevant chemistry and data domain.Next-round precursor, site or topology recommendations and reproducible decision records.

The stage gates prevent an attractive modeled topology from entering broad synthesis without an actual precursor and analytical plan. If the pilot generates several same-mass products, the team does not label the largest chromatographic peak as the desired bicycle without connectivity evidence. Instead, the candidate stays unresolved while the analytical strategy is strengthened or the synthesis is redesigned for higher selectivity.

Deliverables

A Decision Package That Chemistry Teams Can Use

01 · Chemistry Scope & Input Audit

Allowed reaction class, anchor handles, sequences, stereochemistry, known assay values, data gaps and model applicability. The audit records which results are measured, which are computed and which are assumptions.

02 · Site and Precursor Matrix

Candidate cyclization sites with residue numbering, proposed linear precursor sequence, closure bond, ring size and rank rationale. The matrix includes excluded sites and the specific reason for each exclusion.

03 · Bicyclic Connectivity Map

Explicit ring graph, linker-anchor assignment, loop lengths, stereochemistry and plausible alternative isomers. Diagrams use consistent identifiers so that design, synthesis and analytics refer to the same proposed molecule.

04 · Route and Failure-Risk Register

Proposed protection and connection order, anticipated impurities, likelihood and impact of route liabilities, and targeted pilot experiments. We distinguish risks supported by observations from exploratory concerns.

05 · Topology-Confirmation Matrix

Expected mass and diagnostic fragments, required orthogonal analyses, available standards and interpretation thresholds. An unresolved isomeric assignment is reported as such rather than promoted to a confirmed structure.

06 · Pilot Report and Next-Round Design

When experiments are included, conversion and isolation results are compared with predictions; updated candidate ranking and route recommendations accompany a versioned record of assumptions and parameters.

Applications

Where This Service Fits

Closing a New Hit

A target-binding linear hit has several plausible closure positions. We preserve residues implicated in binding, enumerate chemically valid disconnections, and compare an intentionally small panel before ordering an entire macrocycle library. The practical question is which precursor merits the first synthesis attempt and what experimental result would justify expansion.

Moving from Mono- to Bicyclic

A monocyclic lead may benefit from a second constraint, but additional handles can produce alternate bridges or interfere with target binding. We specify the intended graph and linker geometry, assess possible side products and stage a route that permits analytical discrimination among connectivities.

Repairing an Unproductive Route

A previously attempted cyclization gives low conversion or several close peaks. We inspect the original precursor, protection scheme and analytical evidence, then compare new disconnections or connection orders under the same reaction class. A failed route is useful training evidence only when the actual conditions and definition of failure are recorded.

Scaling a Focused Library

Before parallel synthesis, a pilot probes representative ring sizes, residue patterns and linker positions. The resulting observations can calibrate prioritization within the supported chemical space. We set acceptance criteria for conversion and topology identity before scaling, rather than assuming that one successful molecule generalizes to all members.

Scientific Evidence

Why the Closure Site and Reaction Context Matter

A 2026 open-access study on on-resin head-to-tail cyclization assembled 306 standardized peptide cases using a defined diaminonicotinic-acid linker workflow. Within that dataset and reaction setting, a classifier achieved an average 84% cross-validation accuracy for cyclization-site prediction; a 74-peptide experimental set showed 86% agreement with predictions.1 These are published results from the cited research group, not performance claims for this service. Their key transferable lesson is the value of comparable synthesis records, explicit linear precursors and experiments to check predictions.

Published Figure 1 comparing cyclization strategies, alternative closure sites and an ML-guided synthesis workflow
Closure-site alternatives and an experimentally grounded prediction workflow for a defined head-to-tail chemistry.1

Figure 1 shows how distinct disconnections can lead to different synthesis outcomes even when a cyclic sequence is described with the same residue set. It also places computational prioritization after standardized reaction data and before experimental confirmation. Because that study used a particular chemistry and assay definition, its reported accuracy must not be extrapolated to unrelated cysteine linkers or bicyclic routes without local validation.

For bicyclic design, an additional challenge is structural identity. Covalent attachment at several positions can produce architectures with the same gross composition but different connectivity. A reported stepwise chemoenzymatic strategy formed an initial peptide ring followed by a second orthogonal cyclization, illustrating why connection order and compatible reactive handles need to be specified together.2 We use such published routes to frame testable options, not to imply that an untested route will behave identically on a client's sequence.

Our evidence standard therefore asks three separate questions: Did the reaction give a product of the expected mass? Was the desired connectivity distinguished from plausible isomers? Does the characterized product retain the biological and developability profile that motivated cyclization? A positive answer to the first question alone is insufficient for the latter two.

Project Strategy

Computational Prioritization With an Experimental Gate

Define the Applicability Domain

Training examples or published rules are considered only in relation to the chosen linkage, peptide length, sequence composition, protection strategy and experimental protocol. Where the available data are sparse, ranking is qualitative or accompanied by broad uncertainty. We do not present an uncalibrated score as a universal probability of synthesis success.

Use Multiple Evidence Types

Sequence-based signals, simple chemistry filters, conformer ensembles and route review address different failure modes. A low-energy modeled conformation does not prove chemical accessibility, and a high conversion in one analytical assay does not prove the required topology. We record disagreements explicitly and select a discriminating test.

Close the Learning Loop

When the client supplies pilot outcomes, we reconcile the predicted and observed products and update the next design round. Reaction identifiers, precursor structures, linker lot, conditions and analytical methods are retained for traceability. Experimental failures are characterized rather than discarded as uninformative negatives.

Start With a Sequence and a Defined Linkage

Send us a candidate peptide or an unsuccessful cyclization attempt, the allowed linker chemistry and any LC–MS or assay data. We can scope a site-and-topology review, a focused pilot, or an integrated design-to-validation project.

Contact Us →

Related Services

AI-Accelerated Library Design

Generate focused cyclic peptide candidates with target-informed diversity and synthesis constraints.

Explore service →

Oral Cyclic Peptide Lead Optimization

Compare activity, permeability, stability and route feasibility during hit-to-lead decisions.

Explore service →

Experimental Validation

Connect synthesized candidates to structural confirmation, binding, function and property assays where scoped.

Explore service →

References

Selected Scientific Sources

  1. Pan, Y.; Hu, C.; Li, J.; Wan, F.; Hong, X.; Li, C. CycloPepper: a machine learning platform for predicting cyclization outcomes and optimizing synthesis of therapeutic cyclopeptides. Nature Communications 2026, 17, 2803. https://doi.org/10.1038/s41467-026-69441-w. Figure 1 distributed under Creative Commons Attribution 4.0 International License (CC BY 4.0), with cropping of the page margins.
  2. Kobayashi, M.; Onozawa, N.; Matsuda, K.; Wakimoto, T. Chemoenzymatic tandem cyclization for the facile synthesis of bicyclic peptides. Communications Chemistry 2024, 7, 67. https://doi.org/10.1038/s42004-024-01147-w.

Online Inquiry

Submit your project details below, and our team will respond within 24 hours.