LUMIOS – Label using machine in organic samples – A software for dereplication, molecular docking, and combined machine and deep learning

Abstract

LUMIOS, short for Label Using Machine In Organic Samples, is a versatile Python-based software designed to assist professionals and students in organic chemistry with computational exploration of natural products (NPs). Offering user-friendly and NO-CODE application, LUMIOS utilizes three mass spectra categories: catechin, theobromine, and caffeine. Notably, its machine and deep learning models boast accuracy rates exceeding 90%, with a focus on modulating respiratory diseases, particularly through catechin and theobromine. LUMIOS functions by preprocessing molecular information from mass spectra, dereplicating molecules through comparison with a collection of chemical database. The software ensures alignment with the NP class, safeguarding the dereplication process. Annotated structures are then tested in ML and DL models, directing users swiftly to biomacromolecular targets linked to respiratory diseases like asthma and SARS-CoV-2. This enables a streamlined computational screening of the input data.