<p>Solvents are an essential part of liquid chromatography. For categorization, the diversity of solvent properties was captured in several different models. A meaningful description of solvent properties required models with more than two parameters. Showing solvent properties in a graph fosters the reader’s understanding of the diversity of solvent properties, their similarities and dissimilarities. However, there is no universally appropriate solution to prepare such graphs for data sets that comprise more than two parameters. We prepared self-organizing maps for solvents using their descriptors in the solvent parameter model, which were taken from literature. The maps sorted the solvent by similarity and matched solvent classes. In combination with the U-matrix, which captures the change of properties across the map, these maps gave an intuitively comprehensible overview of solvent similarities and dissimilarities. By inspection of the underlying component planes, the grouping of solvents was attributed to an overall framework of solvent properties. The maps are useful both to get an overview of the diversity of the included solvents and to learn how the properties of the solvent parameter model influence the observed solvent differences.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Similarity of solvent properties in self-organizing maps

  • Matthias Guggenberger,
  • Thomas Rosenau,
  • Stefan Böhmdorfer

摘要

Solvents are an essential part of liquid chromatography. For categorization, the diversity of solvent properties was captured in several different models. A meaningful description of solvent properties required models with more than two parameters. Showing solvent properties in a graph fosters the reader’s understanding of the diversity of solvent properties, their similarities and dissimilarities. However, there is no universally appropriate solution to prepare such graphs for data sets that comprise more than two parameters. We prepared self-organizing maps for solvents using their descriptors in the solvent parameter model, which were taken from literature. The maps sorted the solvent by similarity and matched solvent classes. In combination with the U-matrix, which captures the change of properties across the map, these maps gave an intuitively comprehensible overview of solvent similarities and dissimilarities. By inspection of the underlying component planes, the grouping of solvents was attributed to an overall framework of solvent properties. The maps are useful both to get an overview of the diversity of the included solvents and to learn how the properties of the solvent parameter model influence the observed solvent differences.