Move beyond binary affinity. We build and compare protein–PROTAC–E3 ensembles to identify geometries that are stable, cooperative, linker-feasible, and positioned for productive ubiquitination.
Generate ternary-complex ensembles from experimentally solved or modeled protein structures, with restraints that respect both warhead and recruiter binding modes.
Combine structural descriptors and physics-aware scores to rank poses by interface complementarity, linker strain, contact persistence, and ensemble support.
Challenge shortlisted geometries with molecular dynamics instead of relying on one static snapshot.
Compare binary and ternary energetic contributions to identify protein–protein contacts that may stabilize—or penalize—the recruited state.
Evaluate exposed target lysines in the context of modeled ligase assemblies so that pose ranking reflects downstream degradation feasibility.
Translate model evidence into linker, attachment-point, and candidate recommendations that your chemistry and biology teams can test.
Projects can begin with one PROTAC, a linker series, competing E3 recruiters, or an unexplained degradation SAR. The analysis depth is scaled to the decision—not forced into a universal pipeline.
The workflow preserves conformational diversity early, then spends simulation effort only where it can change candidate selection.
Request a Modeling ScopeReview POI and E3 structures, binding-site confidence, ligand poses, attachment vectors, missing regions, protonation states, and any experimental restraints.
Sample protein–protein approaches together with PROTAC conformers, rejecting assemblies that cannot satisfy bond geometry or introduce severe steric conflict.
Cluster poses and integrate interface area, contact quality, linker strain, energetic terms, and recurrence across independent sampling runs.
Run replicate molecular dynamics, examine interface persistence and conformational transitions, and compare ternary behavior against relevant controls.
Map solvent-exposed lysines, assess productive ligase orientation, and deliver a ranked design matrix with proposed validation experiments.
AI-assisted scoring can compress a large pose space, but ternary-complex data remain sparse and system dependent. We therefore report the evidence behind every rank and keep physics-based checks, ensemble agreement, and applicability limits in the same decision table.
Representative output groups designs by confidence and decision value, separating structurally supported candidates from informative experiments that can test competing hypotheses.
Ranked 3D complexes, representative clusters, prepared structures, and clear file annotations.
Linker geometry, interface contacts, energetic terms, stability descriptors, and confidence notes.
Accessible lysines, geometry-based productivity assessment, and competing mechanistic hypotheses.
Candidate tiers, linker or attachment-point changes, controls, and the most informative next experiments.
The authors used molecular dynamics and MM/GBSA to address the three-body nature of PROTAC complexes. In BRD4 BD2 systems, the calculated binding energies correlated with experimental affinity data, and the protocol improved reranking of poses generated by an earlier docking workflow.
Across VHL-recruiting degraders for SMARCA2 and BRD4, ternary-complex affinity and cooperativity correlated with degradation potency and initial degradation rates. The study also used a structural modeling workflow in which total buried interface area agreed with measured ternary binding affinity.
A strong modeling project begins by defining which uncertainty must be reduced and what experimental decision follows.
Preferred inputs are structures or reliable models of the protein of interest and E3 ligase, bound poses or chemical structures for both ligands, linker chemistry, attachment atoms, and any known ternary or degradation data. When a structure is missing, we can first assess whether homology modeling or structure prediction is suitable; that additional uncertainty is recorded explicitly.
Yes. Comparative projects can hold the target warhead constant while testing recruiters, attachment vectors, or linker families. The scope should be designed so that sampling depth remains sufficient for each branch and conclusions are not driven by unequal simulation effort.
No. Ternary-complex formation is necessary but not sufficient. Cellular permeability, target and ligase abundance, ubiquitin-transfer geometry, degradation kinetics, proteasome engagement, and compound exposure can all affect outcome. We position modeling as a prioritization and mechanism tool, not a guarantee of cellular degradation.
AI-assisted scores help rank and cluster large pose ensembles. They are interpreted together with linker feasibility, explicit structural checks, molecular dynamics, uncertainty across independent runs, and experimental data. Final reports distinguish model-supported conclusions from testable hypotheses.
Depending on the decision, useful follow-up data may include ternary binding or cooperativity measurements, target engagement, degradation concentration–response and time-course assays, hook-effect characterization, ubiquitination assays, mutagenesis of proposed interface or lysine sites, and structural studies.
Share the target, E3 ligase, degrader structures, and the decision your team needs to make. CD ComputaBio will propose a fit-for-purpose modeling scope and identify the minimum additional inputs needed.
Submit your project details below, and our team will respond within 24 hours.
Talk to our technical team about your project!
I Want To Talk
