Skip to main content
Fig. 7 | Journal of Cheminformatics

Fig. 7

From: GenUI: interactive and extensible open source software platform for de novo molecular generation and cheminformatics

Fig. 7

A screenshot showing part of the “DrugEx” GUI with a model creation card with (a) DrugEx training parameters and (b) performance overview of a trained DrugEx network. In (a) the fields to define the compound set for the process of fine-tuning the exploitation (‘parent’) recurrent neural network trained on the ZINC data set [60] are shown. In addition, the form provides fields to set the number of learning epochs, training batch size, frequency of performance monitoring and size of the validation set. In (b) the “Performance” tab tracks model performance. It shows values of the loss function on the training set and validation set (top) and the SMILES error rate (bottom) at each specified step of the training process. The performance view is updated according to the chosen monitoring frequency in real time as the model is being trained. Each model also has the “Info” tab which holds the same information as for QSAR models

Back to article page