Imaging Mass Spectrometry (MS) is a maturing technology progressing towards being a routine analytical tool, partially driven by the uptake of high resolving power instrumentation, particularly for small molecule analysis where imaging MS looks set to unlock another dimension in metabolomics. Consequently, more high-quality data is being produced from larger scale studies with significant cohort sizes. To begin bridging the gap between spectra and biological information we have recently developed a high-throughput platform for false-discovery-rate (FDR) controlled molecular annotation of imaging MS. We are now performing inter-study analytics to address some essential questions in spatial metabolomics.
We recruited members of the imaging mass spectrometry community recruitment to submit datasets to our open-access online annotation platform along with some essential sample metadata Every datasets was submitted to our bioinformatic pipeline deployed using big-data technology for high-throughput FDR controlled molecular annotation. All datasets were searched against the Human Metabolomics DataBase for [+H+, +Na+, +K+, or -H-, +Cl- ] adducts in positive and negative mode respectively and annotations reported at an estimated FDR of 0.1 to enable comparison. To mine the thousands of annotations produced we leveraged big-data analytical tools designed for business intelligence of large heterogeneous information streams.
Over 20 laboratories provided more than 300 datasets for annotation (an estimated raw data volume of >60TB). The annotation database includes 18 species, some with multiple disease states, all HR mass analysers, DESI, MALDI and IR-MALDESI ionisation sources. Approximately 6000 unique molecular formulas were annotated across all datasets with a median of ~300 per data-set Untargeted analysis gave us a holistic view over all datasets, enabling us to investigate which molecules are indicative of technology, organism, or tissue type. Alongside high-level queries, comparative analysis permits ‘biomarker discovery’ approaches immediately enhanced with additional information on which other systems discriminative molecules were detected in.