Reference-based screening
Search a project or compound pool from a known ligand using supported molecular, shape, color, or pharmacophore evidence.
Ligand-based screening and analog generation
Explore around known chemistry without losing series context.
Search for molecules related by 2D or 3D evidence, generate interpretable analog hypotheses, and compare each candidate with its parent before sending it into potency, ADMET, selectivity, docking, or synthesis review.
Similarity searchAnalog generationR-group designScaffold hoppingSARChemical space
A connected series workspace
Move from a reference ligand to explainable candidate hypotheses.
MolexIO treats a generated molecule as a proposal with a parent, transformation, and evidence trail. Similarity and generation methods are kept identifiable so chemists can distinguish a close analog, a scaffold-level hypothesis, and a distant chemical-space neighbor.
Interpretable transformations
Retain the operator and transformation label for R-group, bioisosteric, scaffold, growth, or related design hypotheses.
Aligned structural differences
Inspect retained and changed regions with side-by-side depictions and 3D views instead of comparing unrelated orientations.
Series-aware SAR
Keep measured activity distinct from predicted potency and examine how structure and endpoint changes vary across a series.
Downstream evidence
Carry selected candidates into ADMET, off-target, docking, MPO, and route assessment through the shared project identity.
Diversity with control
Balance closeness to the seed with chemical diversity so a ranked list does not become a collection of near duplicates.
From query to candidate set
Use each method for the question it can answer.
- 01
Define the reference
Select a chemically valid molecule and identify the known evidence and design objective.
- 02
Choose the search mode
Use 2D similarity, 3D shape, pharmacophore, or controlled generation according to the available hypothesis.
- 03
Review the series
Inspect transformations, similarity, diversity, properties, and any measured or predicted activity separately.
- 04
Advance evidence
Promote a short list into structure-based, developability, selectivity, and synthesis workflows.
Scientific boundary
Similarity is a search signal, not a biological conclusion.
Activity cliffs, stereochemistry, protonation, assay context, and target mechanism can make similar molecules behave differently. Generated structures may also be unstable, inaccessible, or outside a model's applicability domain. MolexIO preserves method, similarity, transformation, and evidence status so the result remains a hypothesis for expert and experimental review.
Questions before use
Ligand-based design FAQ
Does molecular similarity imply equal activity?
No. It is useful for finding hypotheses, but potency and selectivity must be supported by appropriate evidence.
Can I start without a protein structure?
Yes. A known molecule or series can be sufficient for ligand-based work; structural evidence can be added later.
Are generated analogs automatically synthesizable?
No. They need chemical review and route or supply assessment before synthesis decisions.
Turn a known ligand into a reviewable design space.
Keep the reason for every proposal attached as candidates move into the next decision.