A full-search based tool to rapidly execute your features selection
FAST is the first software of a suite designed to lead computational chemists, labs and companies along the whole QSAR modelling process
Kode Chemoinformatics, the Business Unit of Kode focused on chemoinformatics and AI projects in the chemistry, pharma, food and biotechnology fields, launches FAST, a new product dedicated to relevant molecular descriptors’ selection for QSAR modelling.
Feature selection is a fundamental step in controlling the quality of a machine learning models. Though many solutions have been presented over the past decades, a standard approach to tackle this task is still missing. The quality of each solution can only be evaluated a posteriori and none of the existing approaches shows a theoretical advantage over the others. In QSAR modelling, where the calculation of molecular descriptors may lead to thousands of features, this selection becomes of undisputed relevance.
Our Business Unit of Chemoinformatics faced this issue by developing FAST, a performing software to identify the best subset of molecular descriptors. Designed with a simple and intuitive user interface, FAST is developed to also be integrated into any existing workflow with minimum effort by the operator.
FAST handles the Features Selection through three sequential steps with increasing computational cost and strictness which progressively reduce the dataset size. The sequential approach allows to focus the computational effort of each step on datasets with proper size, making them affordable for the following step. With the time saved, testing a large number of different solutions is therefore accessible even on a standard laptop. Finally, the quality of each set is calculated in terms of prediction performance through one or more machine learning models.
“FAST is the first tool of a complete set of solutions we’re developing to simplifying the job of big data analysis, for those chemists, researchers and molecules developers who have to manage thousands of features in the field of QSAR modelling – declares Alessio Sommovigo, Kode Chemoinformatics BU Manager. – Choosing a suboptimal set of variables may significantly impact on the modelling. I believe our role is to find solutions to spare our clients this kind of problems, by creating easy-to-use tools and easy to integrate tools and products”.
“Our Chemoinformatics BU is Kode’s feather in the cap – adds Marco Calderisi, CEO of Kode – its mission has always been to develop and integrate products and solutions for chemo-informatics, who clearly operate into pretty important fields, such as the environmental one, or in pharma and ecotoxicology areas. Our responsiveness has always distinguished us in the AI market”.
Learn more on Fast