Is there a way to build a classification model using the APF output?
_setAPFparams is in the distribution since 3.6-0; The usage: icm _apf3Dqsar train=trainingSet.sdf activity=LogIC50 table=testSet.sdf Training and test set compounds should be all pre-aligned, for example by aligning training set actives using APF multiple chemical alignment, and then superimposing the test compounds onto aligned actives using APF superposition. Any external alignment method can be used as well. The field containing activity data in the training set SDF is specified by activity= argument. The script also can take alignments in icm multiple object format *.ob, in which case SDFs are only used for input/output of activity data and can be just 2D: icm _apf3Dqsar train=trainingSet.sdf align=trainingSet3Daligned.ob activity=LogIC50 predict=testSet3Daligned.ob table=testSet.sdf The results are written to testSet_predict.sdf output file. Some statistics is reported along the way. If testSet.sdf contains activity (i.e. LogIC50) column like the training set, RMSD and R2 will be
