A prototypic small molecule database for bronchoalveolar lavage-based metabolomics

Abstract

The analysis of bronchoalveolar lavage fluid (BALF) using mass spectrometry-based metabolomics can provide insight into lung diseases, such as asthma. However, the important step of compound identification is hindered by the lack of a small molecule database that is specific for BALF. Here we describe prototypic, small molecule databases derived from human BALF samples (n=117). Human BALF was extracted into lipid and aqueous fractions and analyzed using liquid chromatography mass spectrometry. Following filtering to reduce contaminants and artifacts, the resulting BALF databases (BALF-DBs) contain 11,736 lipid and 658 aqueous compounds. Over 10% of these were found in 100% of samples. Testing the BALF-DBs using nested test sets produced a 99% match rate for lipids and 47% match rate for aqueous molecules. Searching an independent dataset resulted in 45% matching to the lipid BALF-DB compared to<25% when general databases are searched. The BALF-DBs are available for download from MetaboLights. Overall, the BALF-DBs can reduce false positives and improve confidence in compound identification compared to when general databases are used.

A Perspective and Framework for Developing Sample Type Specific Databases for LC/MS-Based Clinical Metabolomics

“…Conversely, searching can be conducted in an iterative fashion across several STSDBs and any compounds that do not match an STSDB can be searched using other DBs and libraries; therefore, the issue with "false negatives", is minimal. Our BAL STSDB [11] also supports an improved identification rate using STSDBs.…”

“…As mentioned, STSDBs should provide an improved compound annotation and ID workflow through an overall quality score (Ov-QS- Figures 3 and 4), which includes a composite compound characteristic quality score (CCC-QS) and relative detection frequency (DF), which counts the number of times a compound has been detected in a given STSDB. A description of these and the overall utility of STSDBs is demonstrated through our prototypic STSDBs, which were recently developed for BAL [11] and HEK293 cells. We are using the term "prototypic" because these STSDBs have not yet been fully curated and the datasets used to populate these DBs are limited.…”

“…Our team has already assembled and tested a prototypic BAL-DB using mouse and human BAL fluid samples [11]. Data processing followed a typical workflow and included additional steps for database assembly (Figures 1 and 3).…”

“…Spectral libraries can contain empirical or in silico derived (i.e., computer generated) spectra. Another strategy entails the use of customized databases that may include some MS/MS spectra, including those that focus on specific sample types [10,11]. For the purpose of this perspective article, we are referring to these as sample type specific databases (STSDBs).While the idea of STSDBs is not new, the full potential of STSDBs remains to be determined, in part because they currently exist as simple repositories of data.…”

Surfactant Lipidomics of Alveolar Lavage Fluid in Mice Based on Ultra-High-Performance Liquid Chromatography Coupled to Hybrid Quadrupole-Exactive Orbitrap Mass Spectrometry

“…Pulmonary surfactant lipids have functions such as maintaining lung stability, protecting lung tissue, preventing alveolar epithelial cell damage and fighting infection [7,8,9]. Scott Walmsley et al described a prototypic small molecule databases for BALF-based metabolomics, including pulmonary surfactant lipids [10].…”