GLP1R PAM Virtual Screen: the Machine Learning Models.

  • GLP1R PAM Virtual Screen: the Machine Learning Models.

Random forest regression and classification models have been tested and assembled in a single ensemble. 

Random forest regression shows an interesting correlation but some undesired overlap between the actives and inactives populations

Random forest regression. Correlation between actual and predicted values.

Activity distribution histogram of predictions by random forest regression algorithms.

Random forest classification models resulted in poor correlations but almost null overlap between the actives and inactives populations

Random forest classification. Correlation between actual and predicted probability of being active.

Activity distribution histogram of predictions by random forest classification algorithms.

Random forest regression and classification ensembles yielded quite acceptable correlations and distribution of populations. 

Random forest regression and classification ensembles. Correlation between actual and predicted values.

Activity distribution histogram of predictions by random forest regression and classification ensembles.