Tailored Mass Spectral Data Exploration Using the SpecXplore Interactive Dashboard

Abstract

Untargeted metabolomics promises comprehensive characterization of small molecules in biological samples. However, the field is hampered by low annotation rates and abstract spectral data. Despite recent advances in computational metabolomics, manual annotations and manual confirmation of in-silico annotations remain important in the field. Here, exploratory data analysis methods for mass spectral data provide overviews, prioritization, and structural hypothesis starting points to researchers facing large quantities of spectral data. In this research, we propose a fluid means of dealing with mass spectral data using specXplore, an interactive Python dashboard providing interactive and complementary visualizations facilitating mass spectral similarity matrix exploration. Specifically, specXplore provides a two-dimensional t-distributed stochastic neighbor embedding embedding as a jumping board for local connectivity exploration using complementary interactive visualizations in the form of partial network drawings, similarity heatmaps, and fragmentation overview maps. SpecXplore makes use of state-of-the-art ms2deepscore pairwise spectral similarities as a quantitative backbone while allowing fast changes of threshold and connectivity limitation settings, providing flexibility in adjusting settings to suit the localized node environment being explored. We believe that specXplore can become an integral part of mass spectral data exploration efforts and assist users in the generation of structural hypotheses for compounds of interest.

Combined LC-MS/MS feature grouping, statistical prioritization, and interactive networking in msFeaST

“…Molecular Networking comprises a combination of data subdivision and exploratory data visualization. Large heterogeneous datasets are subdivided into smaller, more manageable feature groups based on spectral similarity scoring (Nothias et al, 2020;Mildau et al, 2024). Those feature groups are further presented to the user as subnetworks (molecular families) for exploration of relationships across features within them (Nothias et al, 2020).…”

“…Those feature groups are further presented to the user as subnetworks (molecular families) for exploration of relationships across features within them (Nothias et al, 2020). While advantageous as a first analysis step, relationships between subnetworks are lost completely (Olivon et al, 2018;Mildau et al, 2024), parameter setting and optimization are time-consuming and opaque, and further customization of resulting molecular networks in Cytoscape is usually necessary to highlight statistical features for sub-network for feature prioritization (Pakkir Shah et al, 2023).…”

“…This approach provides the basis for meaningful scientific conclusions in group-based comparisons of experimental conditions that can then be further explored in detail (Maleki et al, 2020;Rosato et al, 2018;Chong et al, 2020). To achieve a similar approach in untargeted metabolomics, where genes are replaced by metabolite features with unknown pathway membership, we make use of kmedoids clustering on the pairwise similarity matrix to provide more homologous subdivisions of the data with higher implied structural relationship as in molecular networking (Mildau et al, 2024;Schubert and Rousseeuw, 2021). Feature-set testing is then performed for each cluster using globaltest, providing high power to detect small concordant treatmentspecific effects across cluster members via a single statistical test (Goeman et al, 2004).…”

“…Feature-set testing is then performed for each cluster using globaltest, providing high power to detect small concordant treatmentspecific effects across cluster members via a single statistical test (Goeman et al, 2004). msFeaST draws visualization inspiration from both specXplore and MetGem making use of t-SNE embeddings to represent the entire spectral dataset based on the spectral similarity matrix in one overview (Olivon et al, 2018;Mildau et al, 2024;Maaten, 2008). Furthermore, it draws inspiration from EdgeMaps and specXplore emphasizing topological views of neighborhoods via top-k neighbor edge drawings and group-based color overlays (Mildau et al, 2024;Dork et al, 2011).…”