molDiscovery

molDiscovery performs probabilistic matching of small molecules to mass spectra using an efficient fragmentation algorithm to enable large-scale database searches for compound identification.

Key Features:

• Efficient Fragmentation Algorithm: molDiscovery generates potential fragmentations of small molecules for mass spectrometry analysis using an algorithmic approach that addresses limitations of traditional rule-based chemistry methods.
• Probabilistic Model Learning: It employs a probabilistic model learned from spectral libraries containing tens of thousands of labeled mass spectra to match small molecules to their corresponding spectra and improve search accuracy.
• Enhanced Database Search Capability: The search methodology is reported to be an order of magnitude more efficient than state-of-the-art techniques, enabling searches against databases comprising millions of molecular structures.

Scientific Applications:

• GNPS-scale Identification: Applied to over 8 million spectra from the Global Natural Product Social (GNPS) molecular networking infrastructure, molDiscovery identified nearly six times more unique small molecules compared to previous methods.
• Microbial genomics integration: In microbial datasets integrating mass spectral and genomic data, molDiscovery facilitated discovery of novel biosynthetic gene clusters associated with three families of small molecules.
• Natural product discovery and metabolomics: The increased identification sensitivity supports applications in natural product discovery and metabolomics.

Methodology:

Computational methods explicitly include generation of potential small-molecule fragmentations, a probabilistic model trained on spectral libraries of tens of thousands of labeled mass spectra for molecule–spectrum matching, and an efficient database search algorithm capable of querying databases containing millions of molecular structures.