Does anyone know why I might be experiencing different results with TopKat predictions obtained using TopKat property predictor components [Property Calculators\ADMET\TOPKAT] and the "Toxicity Prediction (TOPKAT)" protocol provided with Pipeline Pilot?
Do these access different versions of the models? Do the components perform some internal processing of the structures that I am overlooking? [The only difference is that I am supplying SMILES for an in-house protocol using the TopKat property predictior components and exactly the same structures written out as an SDF, both of which were generated following standardising the structures using another in-house protocol. I have also explicitly checked that adapting the in-house protocol to read SDF structures still results in differences in the output.]
I am working with Pipeline Pilot version 19.1.0.1964, Build date = Oct 16 2018. I am using the following TopKat component: TOPKAT Rat Oral LD50, TOPKAT Ames Mutagenicity, TOPKAT Daphnia EC50. In the "Toxicity Prediction (TOPKAT)" protocol, I have selected the following models: Ames Mutagenicity (v3.1), Rat Oral LD50 (v3.1), Daphnia EC50 (v3.1).
Thanks in advance for any assistance which can be provided. I appreciate that these are minimal details, but I can try and see if I can reproduce this issue with a public domain compound if necessary.
Best regards,
Richard
-----------
Richard Marchese Robinson
Product Safety QSAR Modelling Expert
Syngenta