|
|
|
|
|
|
|
|
| Stan Young, National Institute of Statistical Sciences |
|
| Stan Young is a statistician who has worked in the pharmaceutical industry for over 30 years. For much of that time he worked on drug discovery problems: the application of space-filling designs for the design of screening sets of compounds and for the design of combinatorial chemistry reaction sets; data mining methods for the analysis of HTS data sets; a new method for pharmacophore mapping. He has won paper and presentation prizes from the ASA and ACS. He a fellow of the American Statistical Association.
|
|
Analysis of HTS data using recursive partitioning
S. Stanley Young, National Institute of Statistical Sciences
High throughput screening (HTS) data is complex; the data sets are large and there are usually active compounds that follow different mechanisms (one QSAR model will not fit all the active compounds). Also, some statistical analysis methods can be complex so that biologists and chemists can be reluctant to jump in even if chemometric help is not available. There is a need for a simple method of analysis that can deal with large data sets and multiple mechanisms. This lecture will present the underlying concepts and analysis strategies of recursive partitioning, RP. We will use the ChemTree® software from Golden Helix, which is particularly easy to use as it was designed from the ground up for the analysis of chemistry data (the lessons learned with ChemTree are largely transferable to other RP codes). Some biological targets have relatively expensive assays and many times resource limitations prevent organizations from scaling up assays to massive HTS for all of their targets. For these targets, sequential screening is a very viable option to progress the target. In sequential screening a modest data set is screened, say 5k to 15k compounds, and then statistical analysis is used to select additional compounds for screening in an iterative optimization process. We will cover the use of RP in sequential screening.
|
|
|
|
|
|
|
|
|
|
|
