Text Mining and Computational Chemistry Reveal Trends in Applications of Laser Desorption/Ionization Techniques to Small Molecules

Abstract

Continued development of laser desorption/ionization (LDI) since its inception in the 1960s has produced an explosion of soft ionization techniques, where ionization is assisted by the physical or chemical properties of a structure or matrix. While many of these techniques have primarily been used to ionize large biomolecules, including proteins, some have recently seen increasing applications to small molecules such as pharmaceuticals. Small molecules pose particular challenges for LDI techniques, including interference from the matrix or support in the low mass range. To investigate trends in the application of soft LDI techniques to small molecules, we combined text mining and computational chemistry, looking specifically at matrix substances, analyte properties, and the research domain. In addition to making visible the history of LDI techniques, the results may inform the choice of method and suggest new avenues of method development. All software and collected data are freely available on GitHub (https://github.com/ReinV/SCOPE), VOSviewer (https://www.vosviewer.com), and OSF (https://osf.io/zkmua/).

Figure 1

(A) VOSviewer term co-occurrence map for MALDI and small molecules and (B) average publication year overlaid on the co-occurrence map. The layout is solely based on the co-occurrences, not the publication year, citations, or any other information. A citation overlay is also available in the VOSviewer file at https://osf.io/gwx54.

Figure 2

(A) Term co-occurrence map for MALDI and peptides or proteins and (B) average publication year overlaid on the co-occurrence map as in Figure 1(B). A citation overlay is also available in the VOSviewer file at https://osf.io/eyfxu.

Figure 3

SCOPE visualization of (A) MALDI, (B) small-molecule MALDI, (C) NIMS, and (D) SALDI, with (E) EI and (F) APCI for comparison. The differences between the MALDI and small-molecule MALDI are subtle but in the expected direction, with more small molecules in the small-molecule MALDI corpus. Each panel also includes the number of unique ChEBIs annotated, the number of papers with text-mined annotations in Europe PMC, and the total number of named-entity recognitions in these papers. Oligosaccharides appear frequently in the NIMS literature with similar log P values around −2.3 (arrow). The SALDI literature produces a small cluster of polymyxins, colistin, and maltoheptaose (arrow). The EI results, dominated by small hydrophobic analytes such as hydrocarbons, and APCI with a “sweet spot” around 550 Da and log P = 7.5 (arrow) were discussed previously and are included only for comparison. An interactive version of this visualization with all queried ionization techniques and analyte classes is available at https://osf.io/rh9bs, with an unfiltered version with the matrix substances included at https://osf.io/qw3ej.