? The development of diagnostic procedures based on advanced proteomic technologies that measure proteins and peptides in clinical cancer studies will require the ability to quantify information from high- dimensional data having thousands of locally correlated measurements. These data can be highly complex with substantial within-group variability as well as non-informative between-group variability. The challenges include detecting and quantifying subtle, yet reproducible, features that define informative proteomic signal. This application is aimed at providing rigorous quantitative methods that increase the power to perform comparative proteomics for current (yet evolving) and upcoming platforms in proteomic research. Therefore, we consider data from a variety of technologies including mass spectrometries, vibrational spectroscopies and multidimensional separation technologies such as chromatograpy and capillary electrophoresis. This application has three primary aims: (1) developing well-defined characterizations of features with the goal of robust detection and accurate alignment of low-abundant peptide signal; (2) normalization so that information extracted from biologically equivalent signals can be quantified for comparison across disease classes; (3) discrimination analysis methods that build on aims 1 and 2, as well as a developing """"""""functional"""""""" discriminant analysis as an alternative for certain technologies in which a simple peak list may ignore subtle or low-abundant signal content. Methods are based on efficiently decomposing signal and variation (via wavelet analysis) for feature detection and normalization as well as employing novel statistical approches to model the variability of dynamic separation processes for feature alignment across different samples. Some of our previous methods have been incorporated into the open-source web-based mass spectrometry analysis tool mslnspect. The methods developed under this application will similarly be translated and implemented via this platform, helping to move it forward and extend its functionality. ? ? ?
| Kundu, Madan G; Harezlak, Jaroslaw; Randolph, Timothy W (2016) Longitudinal Functional Models with Structured Penalties. Stat Modelling 16:114-139 |
| Randolph, Timothy W; Harezlak, Jaroslaw; Feng, Ziding (2012) Structured penalties for functional linear models-partially empirical eigenvectors for regression. Electron J Stat 6:323-353 |
| Holte, S E; Randolph, T W; Ding, J et al. (2012) Efficient use of longitudinal CD4 counts and viral load measures in survival analysis. Stat Med 31:2086-97 |
| Milac, Thomas I; Randolph, Timothy W; Wang, Pei (2012) Analyzing LC-MS/MS data by spectral count and ion abundance: two case studies. Stat Interface 5:75-87 |
| Teckchandani, Anjali; Mulkearns, Erin E; Randolph, Timothy W et al. (2012) The clathrin adaptor Dab2 recruits EH domain scaffold proteins to regulate integrin ?1 endocytosis. Mol Biol Cell 23:2905-16 |
| Yan, Donghui; Wang, Pei; Knudsen, Beatrice S et al. (2012) Statistical Methods for Tissue Array Images - Algorithmic Scoring and Co-training. Ann Appl Stat 6:1280-1305 |
| Brauer, Heather Ann; Libby, Tanya E; Mitchell, Breeana L et al. (2011) Cruciferous vegetable supplementation in a controlled diet study alters the serum peptidome in a GSTM1-genotype dependent manner. Nutr J 10:11 |
| Yu, Xuesong; Randolph, Timothy W; Tang, Hua et al. (2010) Detecting genomic aberrations using products in a multiscale analysis. Biometrics 66:684-93 |
| McLerran, Dale F; Feng, Ziding; Semmes, O John et al. (2008) Signal detection in high-resolution mass spectrometry data. J Proteome Res 7:276-85 |
