An automated workflow by using KNIME Analytical Platform: a case study for modelling and predicting HIV-1 protease inhibitors

Abstract

In this study, we have demonstrated an automated workflow by using KNIME Analytical Platform for modelling and predicting potential HIV-1 protease (HIVP) inhibitors. The workflow has been simplified in three easy steps i.e., 1) retrievethe database of inhibitors for the target disease from ChEMBL website and well-known drug from DrugBank database, 2) generate the descriptors and, 3) select the optimal number of features after machine learning models training. Our results have indicated that the random forest with auto prediction validation method is the most reliable with the best R2 value of 0.9394. Apparently, this workflow can be transformed easily for any other diseases and the quantitative structure-activity relationship (QSAR) model that has been developed can accurately predict in silico how chemical modifications might influence biological behaviour. Overall, the automated workflow which has been presented in this study may significantly reduce the time, cost and efforts needed to design or develop potential HIVP inhibitors.

Study of different data science methods for demand prediction and replenishment forecasting at retail network

“…The workflow consist of many nodes, those are processing the data and transporting results via connections between the nodes. The work in this workflow is organized based on the structure of 4 main stages: a) data collection; b) pre-processing c) model development and training and d) prediction and review of the results (scores) (Ranji et al, 2019). Included methods -logistic regression and random forest.…”