Welcome to Seeker
The Jones research group at Washington State University has been developing a stand-alone application called Seeker that can be used to create recursive partitioning predictive models for 2C9 inhibition. The current beta version of Seeker includes the capability of generating predictions based on values for the nine descriptors used in our recursive partitioning models. For development purposes, we used Chemical Computing Group’s software Molecular Operating Environment (MOE) to generate values for descriptors. A link to details of the nine descriptors is included below.
On this site, we invite users to try out a sample model generated by Seeker. By following the link provided below, a user can input the nine MOE-generated descriptor values for any compound and have Seeker predict whether the compound inhibits 2C9 (defined as having a Ki value above or below 10 μM.)
Also, we include links to a library of 326 compounds (in Excel format) whose structures are given in SMILES string format. This library was used in developing the recursive partitioning models generated by Seeker. This library was partitioned into a 276 compound training set and 50 compound validation set used to test the algorithms. All three spreadsheets are included in the links below and they include the original MOE-generated values for the nine descriptors.
.
Links
Seeker model interface
The entire 326-compound library
The 276-compound training set.
The 50-compound validation set