metabolite_logo

Overview Design Implementation Algorithm Installation Instructions FAQs Acknowledgement

 

The Design Flowchart:

 

Flow chart


 

Input   Two forms of inputs are accepted by MetaboMiner: a peak list and/or a spectra image (currently supports PNG only). The peak list are usually generated from some automatic-peak picking program. It should contain the coordinates of the peaks and optionally, the peak intensity information or some quality indicator (such as S/N values). There are many high-quality NMR-processing software available for this task, including Varian VNMR, Bruker TopSpin , NMRPipe, and Felix, to name a few. The spectra images (.png format) can be obtained after spectral processing using the appropriate program. For more detailed instructions about how to obtain these input, please go to the FAQs page.

Peak processing   Since automatically picked peaks usually contain substantial artifacts as results of baseline distortions, intense solvent lines, ridges, sinc wiggles, etc. It is critical to reduce these artifacts before searching the reference library. Streaks can be recognized by their specific locations and prominent shapes in the NMR spectrum. For TOCSY spectra, real peak signals form a symmetrical square pattern along the diagonal line, which can be exploited to further reduce the artifacts.

Compound identification   There are two steps involved. Firstly, the peaks in the selected reference library are searched against the peaks of the query mixtures. The matched patterns are recorded; Secondly, these patterns are examined for each compound to determine whether they constitute 'minimal signatures' that can uniquely identify the corresponding metabolites. A detailed description of the algorithm is discussed in the Algorithm section.

Spectra annotation   Spectral images can be used to further refine the search result. The peaks of identified compounds will be displayed on the original spectrum. Users can visually annotate the unassigned peaks or edit the reference spectra to match the exprimental data.

Library management   The spectral reference library is the cornerstone of the MetaboMiner. In particular, we created several several sub-libraries that correspond to the metabolite compositions of three common biofluid types (CSF, plasma and urine). Compound identification is much more effective using a library tailered to the biofluids of interests. Researchers studying different biofluids are encouraged to create their own customized library. All library manipulations can be easily achieved by using the GUI.



Last updated:   May 25, 2008
Contact:   Jianguo Xia   jianguox@ualberta.ca   780-4925786