AnnoMe 1.0.0

AnnoMe - Release 1.0.0
AnnoMe is a Python package for the classification of HRMS/MS spectra with the user's aim in mind. Rather than generic compound identification, AnnoMe is designed for targeted natural product discovery: users define what is "relevant" to them, and the tool learns to distinguish compounds of interest from others.

Key Features

• Spectrum Classification — Train multiple classifiers (LDA, NN, SVM, etc.) on annotated MS/MS spectra using cross-validation and majority-voting to assign spectra to user-defined compound classes ("relevant" vs. "other").
• Substructure-Based Filtering — Filter MGF spectral libraries using SMARTS strings, enabling targeted extraction of candidate spectra from large public MS/MS repositories before classification.
• Customizable Classification Workflow — Build and configure end-to-end classification and inference pipelines tailored to specific natural product classes (e.g., prenylated flavonoids, isoflavonoids), supporting training, validation, and inference on novel samples and metabolites.
• Graphical User Interfaces — Two dedicated GUIs for filtering (annome_filtergui) and classification (annome_classificationgui), making the workflow accessible without scripting.
• Public Resource Integration — Automated download of public MS/MS spectral libraries (~11 GB) to serve as training and reference data.

Demonstration
Included demo datasets cover classification of prenylated flavonoids and isoflavonoids using in-house (BOKU University) and public spectral libraries, with inference on plant extracts from Paulownia tomentosa and Glycyrrhiza uralensis.