Raw values retained
Keep the underlying potency, ADMET, selectivity, synthesis, or other supported evidence beside its transformed desirability.
Multi-parameter optimization and what-if analysis
Make the preferences behind a candidate ranking visible.
Translate project objectives into documented desirability functions, compare candidates across heterogeneous evidence, and test how weights, thresholds, and missing values change the decision.
MPO scoringDesirability functionsWhat-if analysisPareto rankingEvidence coverageSensitivity analysis
Transparent decision mathematics
A score should explain the ranking, not end the discussion.
MolexIO keeps raw values, endpoint direction, desirability thresholds, weights, transformed values, uncertainty, and missing-data handling inspectable. Teams can see whether a candidate leads because of balanced evidence, one dominant objective, or an assumption that needs review.
Explicit thresholds
Document whether higher, lower, or a target range is preferred and where unacceptable, transition, and desirable regions begin.
Reviewable weights
Show the relative influence of each objective and prevent an unexplained default from silently defining the program strategy.
Missing-data policy
Distinguish missing from poor evidence and state whether an unavailable endpoint is excluded, penalized, or blocks comparison.
Pareto alternatives
Identify non-dominated candidates that preserve distinct trade-offs before compressing the decision to one score.
What-if scenarios
Change weights and thresholds to reveal rank stability, fragile winners, and the measurement most likely to reduce uncertainty.
Preference, not truthDesirability encodes a documented project decision rule.
Coverage beside scoreIncomplete candidates cannot look fully supported without disclosure.
Scenario preservedWeights and thresholds remain part of the result provenance.
From objectives to a robust shortlist
Challenge the ranking before acting on it.
- 01
Define objectives
Select decision-relevant endpoints and confirm units, direction, comparability, and evidence type.
- 02
Set desirability
Document thresholds, shapes, weights, gates, and missing-data behavior.
- 03
Compare candidates
Inspect raw and transformed values, contributions, coverage, uncertainty, and Pareto status.
- 04
Stress-test
Run what-if scenarios and identify measurements or assumptions that could change the decision.
Scientific boundary
MPO formalizes priorities; it does not make uncertain evidence certain.
Changing a desirability threshold or weight can change the ranking without changing any molecule. Correlated endpoints can be double-counted, missing values can bias comparison, and precise scores can obscure uncertain predictions. MolexIO presents MPO as decision support that requires scientific ownership and appropriate experimental validation.
Questions before use
MPO FAQ
What is a desirability function?
It converts a property value into a normalized preference using documented thresholds and direction; it is not a biological law.
Is the highest score the best drug candidate?
No. It is the leading candidate under the selected objectives, evidence, weights, and policies.
Why run what-if analysis?
It exposes whether the ranking is robust or sensitive to assumptions and missing evidence.
Turn candidate ranking into a reviewable decision model.
Let every stakeholder see the evidence, preferences, trade-offs, and uncertainty behind the shortlist.