Strand has extensive expertise in predictive model building. Strand’s data mining platform has several supervised and unsupervised machine learning algorithms that allow building of predictive models on the fly. The data-mining platform also has feature selection methods and principal component analysis, which help in reducing the dimensionality of the data leading to better and more accurate models.

Strand has deployed this technology and built applications for predicting Pharmacokinetics and Toxicity parameters.

ADME Prediction

• Pharmacokinetics Predictions
• Cardiotoxicity Prediction

Pharmacokinetics Prediction

This platform consists of a curated database containing the pharmacokinetic parameters of 474 drugs and an Artificial Intelligence based set of models that can accurately predict the pharmacokinetic properties of any new compound based on its chemical structure.

Strand has extensively curated the literature and public domain sources to put together a database of the pharmacokinetic properties of drugs that have both succeeded and failed in the market. This database allows sub-structure searching that enables chemists to understand whether their current hits and leads are structurally similar to successful drugs. Based on this understanding a quantitative relationship between the features and each pharmacokinetic parameter has been defined. Strand’s Predictive ADME tool can accurately predict the following pharmacokinetic parameters.

Rate of drug absorption: This is a measure of how quickly an oral dose of a drug gets into the plasma
Bioavailability: This is the fraction of a drug dose that is available in the plasma after being transported across the gut epithelium and having undergone first pass metabolism
Plasma protein binding: This is the fraction of a drug in the plasma that is bound to serum plasma proteins
Volume of distribution: This is a measure of the apparent volume available for a drug to distribute in the body
Half-life in plasma: This is a measure of the elimination half-life of a drug in plasma.

Click here to know more about truPK

Cardiotoxicity Prediction

In silico toxicity prediction can be used to evaluate the potential toxicity profile of a compound even before it is synthesized so as to concentrate expensive resources and efforts on those compounds that are most likely to succeed. Such an approach offers the means to dramatically lower costs and reduce R&D cycle time. Strand has developed models to predict the hERG binding potential of a drug. The hERG channel conducts a repolarising potassium current (Ikr) critical in maintaining normal cardiac rhythm. Impairing this current may lead to fatal cardiac arrhythmias.

Expressed as IC50, which is the concentration of drug required to inhibit Ikr by 50%. Typically binders and non-binders to hERG are defined by the selecting an appropriate IC50 cutoff value. Strand has developed three models for hERG binding that differ in the IC50 cutoff values. These models in increasing order of stringency are:

Model 1: cutoff value is IC50 1µm
Model 2: cutoff value is IC50 5µm
Model 3: cutoff value is IC50 50µm

Click here to know more about truTox