Case Study

Focusing Resources in Hit-to-Lead

Following up hits from primary screening campaigns can be a lengthy and resource intensive process. For programs where multiple hits are found across diverse chemistry, project teams may not have the resource to follow up on all potential lead chemotypes. Synthetic effort therefore needs to be directed towards those chemistries that have the highest overall chance of yielding successful drug candidates.


Following initial screening of over 3,000 compounds selected from a virtual library of over 13,000 molecules, a client identified multiple hits across 13 different chemistries. Due to resource limitations, the project team had to decide where to focus their synthetic effort in order to maximise the probability of identifying chemistries that would yield tractable lead series.


StarDrop enables a project team to define their own criteria for success as a Scoring Profile to prioritise those compounds that have the best overall likelihood of success. The predicted ADME scores for all compounds were compared to assess which chemistries were at high risk of having ADME related problems and which chemistries could be considered to be at low risk.

The project team identified 185 compounds for progression. A large proportion of those selected were located in the area of chemical space predicted to have the best overall balance of ADME properties. However, to minimise risk and gain deeper understanding of SAR, a diverse selection of other chemistries was also included in the set. Initially, these compounds were screened for solubility and human liver microsomal stability. Results from the in vitro screens for the 185 selected compounds agreed with in silico predictions, with compounds in the low risk area of chemical space proving to be the most soluble and stable out of the set.

Applying Probabilistic Scoring

A project team may define a scoring profile that defines the desired outcomes for each property and the relative importance of each of these criteria. In this study, the project team was developing an oral therapy for a chronic treatment against a peripheral (non CNS) target.

Each virtual molecule was scored against these criteria to obtain an overall score which reflected both the likelihood of success of the molecule against the specified criteria, and the measure of confidence in that score with respect to the accuracy and chemical space of the model itself.

The predicted ADME 'heat map' above shows the distribution of these scores with respect to the diversity of the 13,000 compounds in the virtual library. Other analyses provide further information regarding the properties of the chemotypes of interest. For example, the profiles in the figure on the right show the percentage of compounds meeting the criterion for each predicted ADME property. This clearly demonstrates that the compounds in Array 13 are likely to fail to meet multiple criteria, indicating a high risk of failure.

'Snake plots', such as those on the right, provide more detailed information on a compound-by-compound basis. These graphs plot the likelihood of success of each compound (on the y-axis) along with the overall uncertainty in each score (as indicated by error bars). This confirms that Array 13 contains very high risk chemistry, but now it is immediately clear that Array 11 has a significantly higher chance than Array 8 of producing high quality compounds in which all of the property requirements are balanced.

If you'd like more information about the ways StarDrop has been used by our customers, or would like to arrange a demonstration, please contact us. You can also download our product information or try out a free trial version.

Case Study - Chemical space of virtual 13,000 compound library Case Study - Predicted ADME Case Study - <i>In vitro</i> ADME Case Study - Performance against scoring profile Case Study - Snake plots
Latest Forums

Read more >

Popular Downloads

Read more >