Choose leads from a large design series
Compare many linker, warhead, attachment-point, or E3-ligase variants and identify the smallest set that preserves meaningful structural and chemical diversity.
Multi-criteria candidate ranking
Degrader Candidate Prioritization Service for PROTAC Lead Selection
When your degrader pipeline holds dozens of candidates, the cost of advancing the wrong one is high. We score every candidate across structure, potency, selectivity, ADMET, and synthesis so your team advances a defensible shortlist instead of a favorite.
Ternary qualityDC50 & hook effectSelectivityADMET & synthesis
When to prioritize
When This Service Helps You Make the Next Decision
Resolve conflicting assay results
Determine why a compound with strong binding may show weak degradation, poor selectivity, limited permeability, or an unfavorable concentration-response profile.
Decide what to synthesize next
Rank unsynthesized designs using modeled ternary geometry, property predictions, synthetic accessibility, and similarity to measured compounds in the same series.
Prepare a lead-nomination package
Bring structural, biochemical, cellular, ADMET, selectivity, and chemistry evidence into one traceable package for cross-functional review.
Service coverage
Find the Degrader Candidates Most Likely to Succeed
Ternary-complex quality
25%Assess interface complementarity, productive protein orientation, linker strain, accessible ubiquitination geometry, and stability across alternative ternary-complex models. Molecular dynamics evidence can distinguish a plausible docked pose from an interaction network that remains stable over time.
Degradation potency
25%Compare DC50, Dmax, degradation rate, duration, and concentration-response shape rather than relying on one potency value. Hook-effect behavior and assay context are considered so strong performance at one concentration does not dominate the full profile.
Target selectivity
20%Integrate target-family binding, quantitative proteomics, known E3 neo-substrates, and cell-context evidence. Candidates with clean on-target degradation can be separated from compounds that achieve potency through broad or difficult-to-control proteome effects.
ADMET profile
15%Evaluate permeability, solubility, efflux risk, metabolic stability, protein binding, and other exposure-related liabilities relevant to the intended route and tissue. Measured properties are interpreted together because improving one property can worsen another.
Synthesis feasibility
10%Review route length, difficult transformations, linker availability, purification burden, stereochemical complexity, cost, and scalability. This keeps highly ranked designs practical for follow-up chemistry instead of rewarding compounds that are unlikely to be made reproducibly.
Novelty & IP
5%Map chemical-space differentiation across warhead, linker, ligase recruiter, and attachment-point choices. Novelty is treated as a program-specific decision factor and can be raised, lowered, or excluded depending on the project objective.
A tiered shortlist, not a single winner
Candidate rankings are bucketed into decision tiers so chemistry and biology can act on the grouping rather than argue over one decimal place of a composite score.
- AAdvance: strong ternary geometry, robust degradation, clean selectivity, and viable synthesis.
- BOptimize: compelling profile with one correctable liability to address.
- CInform: mechanistically informative controls or boundary designs.
- DDeprioritize: persistent flaws across potency, selectivity, or feasibility.
Image placeholderRecommended asset: a radar/spider chart overlaying two candidates across the six scoring dimensions with a ranked shortlist table beside it.
Flexible project inputs
Start with the Data You Already Have
A prioritization project does not require every candidate to have a complete experimental package. We first separate measured evidence from calculated evidence, identify which gaps could change the ranking, and use modeling only where it adds decision value.
- Accept complete series or uneven candidate datasets
- Retain links to source assays and model versions
- Flag missing, conflicting, and low-confidence values
- Recommend experiments that could change the decision
- Chemical inputs
- Candidate structures or SMILES, stereochemistry, warhead and E3-ligase recruiter identities, linker definitions, attachment points, synthesis status, and related analog series.
- Structural inputs
- Target and E3 structures, binary or ternary-complex models, binding-site information, docking poses, molecular dynamics trajectories, or experimentally determined complex structures when available.
- Biology inputs
- Binding measurements, DC50 and Dmax curves, degradation kinetics, rescue or mechanism controls, cell-line context, target abundance, E3 expression, and phenotypic readouts.
- Developability inputs
- Solubility, permeability, microsomal or hepatocyte stability, plasma protein binding, efflux, cytotoxicity, formulation observations, and early pharmacokinetic data.
- Optional context
- Proteomics, target-family selectivity panels, known off-targets, chemistry constraints, intellectual-property considerations, desired route of administration, tissue goals, and project-specific go/no-go thresholds.
Prioritization workflow
How We Turn Your Candidate Pool into a Clear Shortlist
The workflow gathers heterogeneous evidence, normalizes it into comparable scores, and surfaces uncertainty instead of hiding it.
Request a Prioritization ScopeDefine the decision and weights
Align on the goal—nomination, lead optimization, or mechanistic triage—and set criterion weights with your team.
Assemble candidate evidence
Collect structures, degradation data, selectivity, ADMET, and synthesis inputs for every candidate in the pool.
Generate missing predictions
Fill evidence gaps with modeling—ternary geometry, ADMET, and selectivity predictions—flagged explicitly as computed.
Score and tier the pool
Normalize each criterion, apply weights, and bucket candidates into decision tiers with confidence notes.
Deliver the ranked matrix
Provide the ranked matrix, uncertainty flags, and a validation plan that resolves the top competing hypotheses.
See Why Each Candidate Ranks Where It Does
A single number can conceal that two candidates are close on every criterion except synthesis cost. We expose per-criterion values, normalization rules, evidence sources, and weights so the ranking can be reviewed and adjusted by chemistry, biology, DMPK, and project leadership. The objective is a decision trail your team can challenge—not a black-box score it must accept.
Per-criterionIndividual scores, never just a total
WeightsStated, tunable criterion importance
UncertaintyPredicted vs measured evidence flags
SensitivityRank stability across weight choices
Group Candidates into Clear Action Tiers
Representative output groups candidates by decision value, so chemistry and biology see the advance, optimize, investigate, and deprioritize boundaries clearly. Candidates near a boundary are highlighted for review instead of being separated by an insignificant decimal difference.
AAdvance: strong across all weighted criteria
BOptimize: one correctable liability to address
CInform: controls or boundary designs with value
DDeprioritize: persistent potency or feasibility flaws
Confidence-aware ranking
Keep Uncertainty from Becoming a Hidden Risk
A candidate supported by complete experimental data should not be treated as equivalent to one supported mainly by predictions. Confidence is therefore tracked separately from performance, allowing a promising but uncertain design to remain visible without silently outranking a better-supported lead.
Separate measured and predicted evidence
Every score retains its evidence type, assay or model source, conditions, date, and confidence level. Calculated values can fill a decision gap, but they are not presented as if they were experimental observations.
Test whether the ranking is stable
Weight and threshold sensitivity analyses identify candidates that remain strong across reasonable assumptions. If small changes reverse the shortlist, the result is reported as decision-sensitive rather than definitive.
Resolve conflicting evidence explicitly
Strong ternary geometry, weak cellular degradation, and poor permeability may indicate different bottlenecks. We preserve these conflicts and use them to propose discriminating experiments instead of averaging them into an uninformative score.
Preserve useful chemical diversity
The top tier can include more than the highest-scoring close analogs. Where appropriate, structurally distinct representatives are retained to reduce series risk and give the next experimental round more information.
Decision-ready deliverables
What You Receive from the Prioritization Project
Weighted scoring matrix
Candidate-level values, normalized scores, criterion weights, composite results, data provenance, and sortable rankings for the full pool.
Tiered shortlist
Advance, optimize, investigate, and deprioritize groups with explicit boundaries and a concise rationale for each top candidate.
Candidate evidence cards
Key ternary poses, interaction summaries, linker observations, property profiles, and major strengths or liabilities for priority compounds.
Confidence & sensitivity report
Measured-versus-predicted flags, missing-data map, rank stability across weighting scenarios, and candidates close to tier boundaries.
Validation plan
Recommended assays and experiments selected for their ability to resolve the most important competing candidate hypotheses.
Review session & reusable files
A multidisciplinary results discussion plus presentation-ready figures and editable data tables for internal decision meetings.
Published data
What Studies Reveal About Successful Degrader Selection
Study [1] · Hook effect
Cooperativity, not just affinity, determines productive degradation
Wurz RP, Rui H, Dellamaggiore K, et al. Nature Communications. 2023;14:4177.The authors showed that ternary-complex affinity and cooperativity drive degradation potency and rate, and that excess affinity can suppress degradation via a hook effect. Candidate ranking therefore requires more than potency.
Service implication: prioritization models cooperativity and hook-effect risk so candidates are not advanced on binding affinity alone.
Affinity → cooperativity → potencyOriginal schematic
PROTAC seriesMeasure ternary affinity and cooperativity across variants.
→
Potency & rateRelate ternary attributes to degradation potency and rate.
→
Ranking criteriaPrioritize on cooperative degradation, not affinity alone.
Ternary affinityCooperativityHook effectPotency
Study [2] · Selectivity proteomics
Global proteomics separates clean degraders from promiscuous ones
Donovan KA, Ferguson FM, Bushman JW, et al. Cell. 2018;175(6):1408–1422.Quantitative proteomics across 91 degraders revealed that chemistry and ligase choice produce distinct global degradation profiles, making selectivity a measurable criterion rather than an assumption.
Service implication: selectivity is scored from proteomic and neo-substrate evidence, so promiscuous candidates are flagged even when target potency looks strong.
Degrader set → proteome → selectivityOriginal schematic
Degrader panelProfile degraders by quantitative proteomics.
→
Global profilesMap target and off-target degradation.
→
Selectivity scoreFlag promiscuity as a ranking criterion.
ProteomicsNeo-substratesSelectivityRanking
References
- Wurz RP, Rui H, Dellamaggiore K, et al. Affinity and cooperativity modulate ternary complex formation to drive targeted protein degradation. Nat Commun. 2023;14:4177. https://doi.org/10.1038/s41467-023-39904-5
- Donovan KA, Ferguson FM, Bushman JW, et al. Mapping the Degradable Kinome Provides a Resource for Expedited Degrader Development. Cell. 2018;175(6):1408–1422. https://doi.org/10.1016/j.cell.2018.11.044
Project questions
Common Questions About Degrader Candidate Prioritization
A strong prioritization project starts by defining the decision the ranking must support.
We score ternary-complex quality, degradation potency and hook-effect behavior, target selectivity, ADMET properties, and synthesis feasibility. Weights are tuned to your program goals and the decision at hand.
Weights are agreed with your team at scoping. A potency-driven program weights degradation more heavily, while a selectivity-sensitive program elevates proteomics and off-target considerations.
Yes. Missing data are handled with explicit model predictions and uncertainty flags rather than silently imputed. Ranks distinguish structurally supported candidates from those awaiting confirmatory data.
Yes. High ternary affinity can paradoxically suppress degradation at high concentration. We flag potential hook-effect risk and incorporate it into the prioritization rather than ranking on potency alone.
The framework can be adapted to a focused lead series or a larger virtual design set. The practical project size depends on the depth of structural modeling required, how much experimental evidence is available, and whether missing properties need to be predicted. We can also use a staged approach: rapidly triage the full pool, then apply deeper modeling to the most promising subset.
Yes. The candidate pool may include different E3 recruiters, warheads, linkers, and attachment points. We normalize comparable endpoints while retaining the relevant assay context and structural mechanism. Where cross-series measurements are not directly comparable, the limitation is shown explicitly and the ranking can be reported within series as well as across the full pool.
A ranked candidate matrix with per-criterion scores, weights, confidence notes, a tiered shortlist, and a recommended validation plan for the top candidates.
Start a project
Choose the Right Degraders to Move Forward
Share your candidate structures and any degradation, selectivity, or ADMET data. CD ComputaBio will scope a weighted prioritization and identify the experiments that would sharpen the ranking. Related services: AI PROTAC Ternary Complex Modeling, PROTAC Binding Stability Analysis by MD, PROTAC Bioavailability Prediction.
Online Inquiry
Submit your project details below, and our team will respond within 24 hours.
x
Need help getting the data you need?
Talk to our technical team about your project!
I Want To Talk