Uses the molecule currently selected or drawn in ChemrytIQ as the prediction subject.
Prerequisite: Load The Molecule From ChemrytIQ
Before opening ChemrytIQ-PolyPred, search the molecule in ChemrytIQ by SMILES, InChI, molecule name, or CAS number. Confirm the correct molecule on the ChemrytIQ page, then open the required Chemryt app from that same molecule context so the selected structure is loaded into the app automatically.
What It Does
ChemrytIQ-PolyPred estimates solubility behavior and physical-form signals for the structure currently being reviewed in ChemrytIQ. It helps users understand form-related risk, compare structure-level property signals, and decide when experimental solid-form or solubility follow-up is needed.
Estimates solubility-related behavior for early screening and comparison.
Highlights form-related properties that may affect handling, development, or follow-up experiments.
Quick Tutorial
- Start from the current ChemrytIQ structure and confirm the identity block before prediction.
- Open ChemrytIQ-PolyPred from the child-module list or handoff action.
- Run the solubility and physical-form prediction for the active structure.
- Review predicted solubility behavior, form-related warnings, confidence notes, and any descriptor drivers shown by the page.
- Compare related structures with MCCP or other prioritization views when several candidates are being screened.
- Record the prediction context because solubility and form behavior can depend on salt form, crystal form, purity, and experimental conditions.
Main Areas
| Area | What to review | When to use it |
|---|---|---|
| Structure representation | Current ChemrytIQ molecule, salt/form assumptions, and descriptor handoff state. | Use before prediction. |
| Solubility panel | Predicted solubility behavior, units where available, uncertainty, and source/model notes. | Treat as screening guidance. |
| Physical-form panel | Form-related signals and warnings that may affect development or handling. | Use to plan follow-up characterization. |
| Comparison | Candidate rankings, property tradeoffs, and downstream prioritization. | Use when reviewing a related compound set. |
ML Model / Computation Used
| Model or method | What it predicts | Implementation details |
|---|---|---|
| PolyPred solubility ONNX models | Aqueous logS, hydration free energy, single-solvent logS, and binary-solvent mixture logS. | Task models use ONNX plus preprocess.npz. Feature vectors combine Morgan fingerprints with RDKit descriptors; solvent tasks concatenate solute/solvent features and scalar conditions such as temperature or solvent fraction. |
| COSMO sigma-profile compact Random Forest | 51-bin COSMO-RS sigma-profile estimate for physical-form/solubility context. | Merged ONNX RandomForestRegressor graph trained from SELFIES-derived bag-of-token features; compact deploy variant reports overall R2 about 0.49 and MAE about 1.11 on held-out data. |
Good Practice
Solubility and physical-form prediction is screening support only. Validate important form, salt, polymorph, and solubility decisions with measured experimental data.
Reference Used
This Tutorial page mirrors the ChemrytIQ reference module: ChemrytIQ-PolyPred.