A Quantitative Structure-Activity Relationship (QSAR) model can be created using Machine Learning (ML) methods to forecast biological responses by forming a mathematical connection between a molecule’s structure and its activity. This process includes applying ML algorithms to a dataset comprising chemical compounds along with their known biological activities, and then utilizing the established relationship to estimate the activity of new, untested compounds. Orange is an open-source toolkit for data visualization, machine learning, and data mining. It provides a visual programming interface for interactive qualitative data analysis and exploration which is developed by the University of Ljubljana. Orange serves as a visual programming platform for data science and machine learning endeavours. Prediction algorithms based on machine learning, such as random forest and SVM, may be employed and validated using methods like leave-one-out (LOO), random sampling, and cross-validation to assess the biological response. In this workshop, a brief overview of machine learning methods and algorithms is provided, which includes supervised, unsupervised, and reinforcement learning, accompanied by examples. Furthermore, the application of machine learning in QSAR modelling is discussed to provide insights into predicting the biological responses based on a diverse range of physiochemical descriptors.

The aim is to establish a reliable model that can effectively forecast the biological activity of novel, untested compounds, which could expedite drug discovery and various other applications.

• To create a Quantitative Structure-Activity Relationship (QSAR) model utilizing machine learning (ML) techniques to forecast a biological response based on the structure of a chemical compound.
• To identify pertinent molecular descriptors, choosing suitable machine learning algorithms, and thoroughly validating the model’s predictive performance.