ADMET Predictor® User Manual
ADMET Predictor® User Manual
Breadcrumbs

7.7.2. Global Model Predictions

7.7.2. Global Model Predictions

  1. The input file provided in this tutorial already contains predictions from the global QSAR model GLB_RLM_CLint. Descriptors have been precalculated as well. You could use the CYP_RLM_CLint model in ADMET Predictor to run this tutorial if you have a license for Metabolism module, but will see different results if you do. 

  2. Compare the predicted RLM CLint log(GLB_RLM_CLint) with the experimental CLint log(OBS_RLM_CLint) by plotting them on a scatter plot.

image-20260602-233250.png

  1. Click on the Chart Properties (1) icon in the toolbar along the top of Chart 2 or some other 2D scatter plot. Clicking on (2) in the chart toolbar will cause the linear fit statistics to appear at the top of the plot. The GLB_RLM_CLint predictions are good: the RMSE for the 51 compounds with CLint data is 0.419 log units (about 2.5-fold) and the R2 is 0.605.

  2. image-20260602-234417.png

    The next thing to do is to build a local ANNE regression for the RLM CLint using only information from the local data set without taking advantage of information contained in the GLB_RLM_CLint predictions.