Percepta Predictors


Click to view the image series

Get an in-depth understanding of structure-property relationships with ACD/Labs' fully powered predictive models for physiochemical and ADME properties, and toxicity endpoints.

The predictors are a collection of expert modules built for comprehensive evaluation of structure-property relationships. The advanced algorithms that power these predictions have been built on the combined strengths of ACD/Labs and Pharma Algorithms (the companies merged in 2009). In the Percepta platform we have brought together our joint experience and expertise to provide the best from all models, and the fastest, most accurate predictions in almost two decades of development.

Predict properties from simple structure input
A variety of input parameters (name, 2D structure, or SMILES string) provide predicted results.

Evaluate predictions in single structure or spreadsheet view
With each module you have the ability to view details of the prediction for a single structure, and interactively adjust parameters of interest (where applicable). Each module offers different prediction-specific tools/information including:

  • Colour coded mapping on the structure to highlight atomic/substructure contributions
  • Interactivity with structures to assess contributions from different structural elements
  • Graphs showing the effect of pH
  • Calculation protocols

Spreadsheet view offers the additional capability to view predictions from all licensed modules in one screen, and a number of graphing, sorting, and filtering tools to rank compounds and aid evaluation.

Assess predictions using reliability index and similar structures
Each module provides either a probability or reliability index for the prediction. Use these numbers to assess confidence in the predicted result. Display of up to 5 of the most similar values from within the training set (with literature data and references) offer the ability to judge the relevance of the training set to your chemical space.

Train models with in-house experimental data
To better reflect proprietary chemical space and improve prediction accuracy, a number of the prediction modules (see table below) offer the ability for you to train the prediction model with experimental data. Training is user-friendly and may be switched on, off, or certain training sets used for different prediction, giving you full control.

Apply different models
We offer a number of different models for the prediction of logP and pKa—ACD/Labs classic, ACD/Labs GALAS, and, for the first time, a consensus model for logP prediction. Evaluate and investigate predictions from the different models and use the most appropriate for the calculation of ADME properties.

Available Modules

We offer a broad collection of powerful prediction modules for physicochemical properties, ADME outcomes, and toxicity endpoints. Customize your own suite of tools by combining any number of modules below.

PhysChem (click to expand)

  • Aqueous Solubility*
    • Calculate pH dependent aqueous solubility, intrinsic solubility, and solubility of the chemical dissolved in pure (unbuffered) water at 25°C and zero ionic strength
    • View results in tabular/graphical form as a function of pH with references to experimental research
    • View tabular/graphical representation of % ionic species at given pH values
    • View and select appropriate tautomeric forms
    • Include melting point data for solids
    • Train the model with experimental values to improve predictions for proprietary chemical space
  • Boiling Point
    • Predict the vapor pressure of organic compounds as a function of temperature, boiling point temperature, or pressure
    • Estimate boiling point at atmospheric pressure
    • Calculate enthalpy of vaporization at 760 mmHg
    • Calculate flash point
  • LogD*
    • Calculate the distribution coefficient (logD) at any pH (pH 0–14), with or without ion-pair partitioning
    • View results in tabular/graphical form with % ionic species at given pH values
    • Calculate the % dominant form in aqueous and organic phases
    • Calculate BCF and Koc at any given pH
    • Train the model with experimental values to improve predictions for proprietary chemical space
  • LogP*
    • Calculate the octanol-water partition coefficient for a wide range of neutral compounds under standard conditions, at 25°C
    • Calculations are provided with 95% confidence intervals, or reliability Index (RI)
    • Review bioconcentration factor (BCF) and the adsorption coefficient (Koc)
    • Evaluate Rule-of-5 compliance
    • Train the model with experimental values to improve predictions for proprietary chemical space
  • pKa*
    • Calculate accurate acid-base ionization constants (pKa values) under 25°C and zero ionic strength in aqueous solutions for organic structures. Specific constants calculated:
      • Apparent
      • Approximate
      • Exact microscopic
      • Single
    • Calculations are provided with 95% confidence intervals and a detailed report on how it has been carried out (including Hammett-type equation(s), substituent constants, and literature references where available); or with Reliability Index (RI) of the calculation and display of the 5 most similar structures for each ionizable site in the training set.
    • View tabular/graphical representations of Net charge vs. pH (with breakdown of % ionic species at any given pH)
    • Train the model with experimental values to improve predictions for proprietary chemical space
  • Sigma
    • Calculate the Hammett electronic substituent constant (s, sigma) for selected fragments of a molecule, or substituents
    • Calculate only inductive and resonance sigma constants (option)
  • Absolv
    • Calculate solvation-associated properties from Abraham-type equations
  • Adsorption Coefficient (Koc)
  • Bio-Concentration Factor
  • Density
  • Freely Rotatable Bonds
  • H-Bond Donors and Acceptors
  • Index of Refraction
  • Molar Refractivity
  • Molar Volume
  • Molecular Weight
  • Parachor
  • Polar Surface Area
  • Polarizability
  • Rule-of-5
  • Surface Tension

ADME (click to expand)

Tox (click to expand)

*Trainable modules.

For more information or to request a demo, contact us.