BioPredict has experience and proprietary tools to help with the design and interpretation of HTS data. 

i.  Selection of Compounds for Initial Diversity Screening.

 We have developed proprietary that allow rapid profiling of very large compound libraries for diversity and drug-likeness.  The application of these tools provides a description of the coverage of compound classes in the initial compound library as a function of subset size. This analysis is a useful decision tool when considering the trade-off coverage vs. the cost of the initial screen.

Targets of pharmaceutical interest increasingly fall into protein families for which there is prior knowledge at the medicinal chemistry level.  BioPredict is able to enrich compound selection in these cases  by the use of training-learning methods on known actives within the class and using them to  identify compounds in the library that are likely to be active.  A comprehensive set of proprietary learning methods has been developed (see Technologies)  to identify novel actives not necessarily within the same chemical class.

ii. Interpretation of HTS Results and the Design of Follow-on Screens

BioPredict has extensive Hit-to-Lead capabilities, critical in the identification of lead candidates with favorable potency and predicted ADME/TOX profilesas well as  multiple positive and negative examples to establish SAR.  

Interpretation of HTS results.  BioPredict has developed an extensive set of data analysis tools specifically for the interpretation of high throughput screening data.  BioPredict applies these tools HTS data to:

·        Identify potential leads as clusters of active compounds;

·        Analyze close negatives for each cluster to construct an initial SAR;

·        Identify potential false positives as positives surrounded by close negatives;

·        Identify potential false negatives as negatives surrounded by close positives;

·        Classify isolated positives as worth pursuing or not based on the number of
   similar compounds tested in the screen ;

·        Interpret the modes of binding of potential leads when the structure of the target
   protein or of a close homolog is available.
 

Design of follow-on Screens.  BioPredict has developed an extensive set of data-mining tools to select compounds for follow-on screens.   This list is comprised not only of untested compounds, but also potential false positives and negatives from the initial screen whenever they are critical to an SAR.   In the past BioPredict scientists have applied computational methods (see Technologies) to construct well-designed second screens that have:

·        Doubled the total number of actives

·        Doubled the number of actives in leading active classes

·        Significantly increased activity within leading classes

·        Recovered false negatives as true positives for SAR

·        Identified false positives as true negatives for SAR

 

Progressive Screening.   An alternative strategy to screening a large and fixed number of compounds by diversity is to analyze the data for the first 20%  of compounds screened and to bias the next 20% to include predicted actives based on the analysis of the initial phase. Such a strategy intrinsically creates flexibility to respond to the data as it emerges, maximizing return while minimizing time and expense.