Overview Design Implementation Algorithm Installation Instructions FAQs Acknowledgement


Library creation and curation

Two widely used 2D NMR methods are supported: 1H-1H total correlation spectroscopy (TOCSY) and 1H-13C heteronuclear single quantum correlation (HSQC). The spectral information were obtained from public resources (Human Metabolome Database, BioMagRasBank, and Magnetic Resonance Metabolomics Database). These data were manually curated and cataloged into different biofluid libraries according to their respective metabolite compositions. Since this information was obtained using a variety of analytical techniques not limited to NMR, we also created a series of "common" libraries based on the compound lists identified from a large number of 1D 1H spectra of each type of biofluid.

The libraries are stored as XML files and can be easily edited or extended via the graphical user interfaces (GUI) provided by MetaboMiner.


Summary of spectral reference libraries
NMR Exp. Biofluid All Common
       TOCSY   Generic 223 60
  CSF 82 39
  Plasma 137 48
  Urine 95 48
        HSQC   Generic 502 61
  CSF 146 39
  Plasma 287 48
  Urine 185 61


Software implementation

The software was written in the JavaTM programming language. In particular, the graphical user interfaces were implemented using Java Swing technology. The spectra visualization and manipulation functions were implemented using the JGraph library (Java open source graph visualization library).

Automatic peak picking programs usually pick up large numbers of peaks (~1,000 peaks for HSQC spectra and 2,000~3,000 for TOCSY spectra). Our tests show that MetaboMiner works well for up to ~5,000 peaks. Users are suggested to use a low cut-off threshold during the process of automatic peak picking.




Last updated:   May 25, 2008
Contact:   Jianguo Xia   780-4925786