KPLS Definition (.kpl)
The naming convention for KPLS model definition (.kpl) files is:
<data file name>.kpl
The KPLS file contains all data used for training the models and the optimal parameters for each member model in a KPLS ensemble. The file consists of the following blocks, within each of which values are separated by tabs.
Program identifier
# ADMET Modeler(TM) <version> KPLS <number of members>
#=====================================================
All training data
AllTrainingData <#points> <#descriptors> <x1> <x2> … <xN>…
InputNames <name1> <name2> ...<nameN>
InputShift <shift1> <shift2> …<shiftN>
InputScale <scale1> <scale2> …<scaleN>
InputMin <min1> <min2> … <minN>
InputMax <max1> <max2> … <maxN>
OutputName <outputname>
OutputShift <outputshift>
OutputScale <outputscale>
OutputMin <outputmin>
OutputMax <outputmax>
#==================================================
Member model
TrainingPoints <index1> <index2> … <indexm>
Kernel <kernel type> <kernel parameter>
LV <number of latent variables>
Alpha <value1> <value2> … <valuem>
Kappa <value1> <value2> … <valuem>
TrainingStats <value1> <value2> … <value8>
VerifyStats <value1> <value2> … <value8>
TestStats <value1> <value2> … <value8>
Merit <value>
#--------------------------------------------------------------------
…
The Member model section is repeated for every member model. The training points indices are zero-based, so the number 0 in the All training data section refers to the first record.
Here is an example KPLS model file, ellipses indicate omitted lines and entries.
# ADMET Modeler(TM) 1.4.4 KPLS 32
# ===========================================================================
AllTrainingData 195 85
-2.439120 -2.3775875 -2.358277 -0.824323 -0.907141 …
-2.305001 -2.1907609 -2.205161 -0.824323 -0.907141 …
…
1.8547987 1.82601792 2.0820785 1.1629987 -0.907141 …
InputNames MWt MolVol N_Carbon N_Ntrgen N_Oxygen …
InputShift 399.274395 386.249196 24.401929 2.829581 1.2315119
InputScale 104.585795 112.403498 6.5310076 1.006379 1.3575727
InputMin 144.177 119 9 -1.747733E-16 -6.808789E-17
InputMax 628.693 609 38 6 5
OutputName pIC50
OutputShift -2.2399437
OutputScale 1.5207652
OutputMin -4.9435
OutputMax 2.0969
# =============================================================================
TrainingPoints 0 2 3 5 6 8 9 10 11 …
Kernel RBF 15.443905
LV 6
Alpha -1.869357 -1.397004 -0.579559 -7.398603E-03 0.148739
Kappa -4.186190E-03 -1.371308E-02 9.0708861E-03 5.016250E-03 -1.843350E-03
TrainingStats 0.230 0.175 0.894 0.980 -977 0.968 -0.057 -977
VerifyStats 0.868 0.678 0.810 0.705 -977 0.956 -0.102 -977
TestStats 0.870 0.661 0.647 0.627 -977 0.808 -0.468 -977
Merit 1.243114
# --------------------------------------------------------------------------------