MS2AI

MS2AI automates repurposing of public peptide liquid chromatography-mass spectrometry (LC-MS) data into standardized datasets for machine learning applications in proteomics.

Key Features:

• Automated Data Retrieval: Retrieves mass spectrometry data from in-house sources and public databases such as PRIDE.
• Data Standardization: Standardizes raw LC-MS data into machine-learning–suitable structured formats including MS1 and MS2 spectra and peptide identifications.
• Dataset Aggregation for Balanced Training: Aggregates extensive LC-MS and mass spectrometry datasets to mitigate limited training sample sizes.
• Information-rich Data Representations: Defines information-rich data representations required for peptide identification tasks.
• Benchmarking Support: Facilitates benchmarking of machine learning methods tailored to LC-MS problems.
• Machine Learning Integration: Includes a convolutional neural network (CNN) example to identify oxidized peptides.

Scientific Applications:

• Proteomics Research: Supports protein identification and quantification in proteomics through machine learning-enabled analysis of LC-MS data.
• Peptide Identification: Processes and analyzes peptide identifications for studies of post-translational modifications and peptide-level analyses.

Methodology:

Retrieves raw LC-MS data from in-house sources and PRIDE, standardizes the data into structured ML-ready formats including MS1/MS2 spectra and peptide identifications, aggregates datasets to enable large-scale and balanced machine learning applications, and demonstrates machine learning application via a convolutional neural network for oxidized peptide identification.