The application of ultrahigh-field Fourier transform ion cyclotron resonance mass
spectrometry (FTICR-MS) technology to identify and quantify metabolomics data is
relatively new. An important feature of the FTICR-MS metabolomics data is the high
percentage of missing values. In this thesis, missing value analysis showed that the
missing value percentages were up to 50% and the control treatment, NaOH.ww, had
the highest missing value percentage among the treatments in the aqueous FTICRMS
sets. A simulation study was done for the FTICR-MS data to compare selection
methods, the Kruskal-Wallis test and the MTP and Limma functions in Bioconductor,
an open source project to facilitate the analysis of high-throughput data. The study
showed that MTP was sensitive to variations among treatments, while the Kruskal-
Wallis test was relatively conservative in detecting variations. As a result, MTP
had a much higher false positive rate than Kruskal-Wallis test. The performance of
Limma for sensitivity and false positive rate was between the Kruskal-Wallis test and
MTP. Data sets with missing values were also simulated to assess the performance of imputation methods. Study showed that variances among treatments diminished
or disappeared after imputations, but no new differentially expressed masses were
created. This gave us confidence in using imputation methods. Summary of analysis
results of some of the frogSCOPE data sets was given in the last chapter as an
illustration. / Graduate
| Identifer | oai:union.ndltd.org:uvic.ca/oai:dspace.library.uvic.ca:1828/3566 |
| Date | 09 September 2011 |
| Creators | Lu, Linghong |
| Contributors | Lesperance, Mary |
| Source Sets | University of Victoria |
| Language | English |
| Detected Language | English |
| Type | Thesis |
| Rights | Available to the World Wide Web |
Page generated in 0.0019 seconds