MFTA Models in Virtual Screening

1 MFTA-based Virtual Screening of Compound Databases 

In general, the virtual screening procedure for the database of available compounds involves the following steps  

Normalization of the structure data set e.g., removal of explicit hydrogen atoms, standard representation of nitro, azido and similar groups, deionization , and removal of duplicate structures). 

Preliminary filtering by structural class/scaffold (if desired for the extrapolation or the targeted exploration purposes, wider chemical domain can be covered compared to the training set of a model). 

Prediction of bioactivity endpoints by means of the MFTA models. Post-filtering of structures that fall outside the model applicability domain and have unreliable predicted values. 

---