SVD- -based Functional ANOVA For based Functional ANOVA For SVD - PowerPoint PPT Presentation

SVD- -based Functional ANOVA For based Functional ANOVA For SVD Measurement Evaluation of MALDI- - Measurement Evaluation of MALDI TOF Mass Spectrometry of Polymers * TOF Mass Spectrometry of Polymers * Z.Q. John Lu (John.lu@nist.gov) Statistical Engineering Division, ITL National Institute of Standards and Technology *Acknowledgement: Presentation based in part on collaboration with Charles Guttman (NIST), Stephanie Wetzel (NIST), Jennifer Huckett (Iowa State).

I. Background I. Background 1. Statistics in Biology : beginning of mathematical statistics (R A Fisher); statistical genetics; genomics (microarrays), proteomics, system biology (Nobert Wiener). 2. High-throughput experiments and modern metrology: large data (many variables, high p), and unfortunately not many samples (low n)! 5/28/04 Interface2004 2

A G E P Box view of statistical learning? A G E P Box view of statistical learning? H. Kitano 2002, Science, System Biology H. Kitano 2002, Science, System Biology 3

3. Biomarker hunting from serum mass 3. Biomarker hunting from serum mass spectra spectra • Qu, Y, B. Adam, et al (2003): Biometrics, 59, 143- 151. – P=48,538, n=248 (167 cancer patients and 81 controls) – Test samples= 45 • Claim: data too large for PCA or SVD – Opts to use wavelet transform, K-L information criterion for variable selections, then apply Fisher’s discriminant analysis, evaluating classifier: sensitivity=98%, specficity=99%. 5/28/04 Interface2004 4

SELDI Serum Protein Profile Analysis- - Prostate Cancer Prostate Cancer SELDI Serum Protein Profile Analysis Bao- Bao -ling Adam et al 2004 Interface ling Adam et al 2004 Interface 7 5 5 0 N1 2 5 0 7 5 5 0 N2 2 5 0 7 5 5 0 N3 2 5 0 7 5 5 0 N4 2 5 0 7 5 5 0 N5 2 5 0 7 5 5 0 N6 2 5 0 7 5 5 0 B1 2 5 0 7 5 5 0 B2 2 5 0 7 5 5 0 B3 2 5 0 7 5 5 0 C1 2 5 0 7 5 5 0 C2 2 5 0 7 5 5 0 C3 2 5 0 7 5 5 0 C4 2 5 0 6 0 4 0 C5 2 0 0 7 5 5 0 C6 2 5 0 5/28/04 Interface2004 5 2000 6000 10000 2000 6000 10000

II. Statistical and Metrological Issues II. Statistical and Metrological Issues • High-dimensional pattern recognition and regression modeling (prediction) – What are the underlying assumptions for pattern recognitions such as the simple hyperplane classifer? – Hidden and intrinsic low-dimensional predictors and classifying variables? – Need robust and data and user-friendly, algorithmic like, and fast, scale well • Repeatability and reproducibility controls (Baggerly et al 2004, Bioinformatics) • Experimental design to validate conclusions: sample size requirements 5/28/04 Interface2004 6

Biostatistics to the rescue? to the rescue? Biostatistics • Boguski and McIntosh 2003: “The analysis of proteomics data is currently informal and relies heavily on expert opinion”. • Recommend: better study design to avoid confounding factors • Study of biological variability and reproducbility 5/28/04 Interface2004 7

III. MALDI- -TOF Mass Spectrometry of TOF Mass Spectrometry of III. MALDI Synthetic Polymers Synthetic Polymers 1. The molecular mass (weight) distribution (MMD) of synthetic polymer is studied using MALDI-TOF MS. The MWDs and the moments derived from MALDI are compared with the values determined by traditional methods including size exclusion chromatography, light scattering, and osmometry. 2. Evaluate effects from experimental design of various instrument settings and sample preparation. 5/28/04 Interface2004 8

6 5 6 6 5 6 6 5 5 6 6 6 6 5 6 6 6 6 5 6 6 5 5 6 6 6 5 5 6 6 5 5 6 6 6 7 7 6 7 4 3 1 1 5 4 7 3 5 8 3 2 4 2 1 2 7 7 4 4 4 2 1 3 3 8 1 3 3 8 8 8 2 4 6 4 1 5 5 6 6 7 3 8 3 1 6 1 1 6 6 6 5 3 3 3 4 4 2 5 7 5 4 4 3 3 4 7 7 7 4 1 2 7 3 3 2 8 8 7 7 2 6 1 6 2 5 4 2 4 4 8 5 3 8 8 4 8 6 2 8 8 5 5 2 8 3 3 5 8 3 4 6 6 1 7 4 2 7 3 7 2 7 7 2 8 8 3 4 4 6 6 7 7 8 8 2 2 3 4 8 1 3 3 2 1 5 1 4 3 2 1 5 5 5 5 2 3 3 4 4 7 5 2 7 4 7 8 1 2 1 4 5 1 3 3 5 4 7 2 6 2 6 4 2 4 4 7 8 6 5 7 4 2 3 6 4 1 2 7 7 7 3 3 7 5 4 5 2 1 4 5 3 3 1 7 2 4 8 4 1 2 7 3 2 3 6 2 8 5 5 3 8 3 3 1 7 5 7 1 3 8 1 2 6 8 6 6 1 2 1 7 8 3 4 5 1 2 5 1 5 4 7 6 7 1 2 1 6 7 4 1 4 1 3 2 6 2 6 3 6 7 7 2 8 4 7 7 4 4 7 6 5 5 8 5 6 4 2 3 7 3 8 1 7 3 5 1 2 2 8 1 6 5 4 5 7 7 3 4 6 2 3 4 1 7 4 1 7 2 4 8 2 2 1 2 5 3 7 5 3 3 3 3 8 8 4 5 8 1 2 7 1 2 3 8 5 1 4 3 4 8 6 6 5 2 3 8 1 6 3 1 7 7 5 5 5 1 2 1 2 3 1 2 7 1 2 4 5 3 5 1 7 2 7 4 2 5 3 4 4 6 3 4 5 1 6 1 4 8 4 5 8 8 7 1 6 6 3 1 6 5 7 6 2 2 5 3 3 2 8 4 8 8 3 3 5 7 4 3 2 3 6 6 4 2 4 4 3 2 5 6 3 6 7 3 7 7 1 2 5 4 1 7 4 4 8 6 8 3 8 1 7 6 6 1 4 7 7 2 7 8 1 4 7 5 2 5 4 2 5 4 8 4 1 6 5 3 7 2 4 6 6 1 1 2 5 6 3 4 5 8 1 8 7 3 3 3 8 2 3 5 3 4 3 1 1 2 2 2 1 2 5 2 3 4 6 4 5 8 5 7 1 7 3 6 1 7 2 2 2 7 3 7 5 4 4 2 4 4 1 5 7 2 2 1 1 8 4 2 7 1 1 1 7 5 1 5 6 6 2 7 3 8 3 6 1 7 1 4 4 1 7 7 6 1 8 3 5 1 5 6 4 1 5 4 8 1 4 5 4 8 1 1 7 3 2 5 4 7 3 7 1 1 5 7 8 1 7 1 1 7 3 1 6 7 8 7 2 1 5 2 6 8 6 6 1 7 1 6 5 5 2 5 3 8 7 2 2 1 5 2 5 1 2 6 6 2 6 5 3 5 3 5 7 1 1 5 7 7 4 1 1 7 4 2 4 3 3 4 2 7 6 6 1 4 1 5 4 1 1 5 1 8 7 7 1 8 2 1 7 1 6 4 4 2 6 6 1 1 4 3 8 2 2 2 7 2 2 6 3 1 8 2 6 4 4 3 2 5 1 3 3 1 2 5 2 2 3 7 5 7 4 1 5 3 7 5 7 7 7 8 8 6 3 1 6 2 5 1 1 5 2 2 6 8 8 5 3 1 1 2 4 3 2 3 6 5 2 7 8 2 2 6 6 1 7 4 1 6 5 5 4 8 4 7 1 6 3 1 1 7 7 4 2 1 2 1 3 3 2 3 2 3 5 5 4 7 7 3 4 3 1 4 7 2 2 6 5 4 2 1 3 6 6 6 4 7 5 7 5 7 5 6 8 2 8 1 5 8 4 3 8 7 4 3 2 1 1 1 6 4 2 5 3 7 2 3 7 5 4 2 7 5 4 3 4 4 8 4 4 4 7 3 7 6 1 8 5 8 6 2 1 1 1 3 1 3 7 8 2 2 4 4 4 4 4 1 5 3 2 7 6 6 4 1 5 5 5 2 3 3 1 7 7 2 7 4 6 6 2 2 5 3 5 3 1 2 8 8 1 3 8 1 3 4 6 4 7 2 7 2 6 6 6 4 4 1 1 5 3 5 5 7 7 6 8 7 2 7 8 6 6 3 1 6 3 5 3 4 2 5 1 5 5 5 1 8 7 4 8 7 4 3 2 2 7 6 5 7 4 4 6 8 2 3 5 1 1 6 6 3 7 1 2 2 7 3 8 4 5 5 1 6 6 3 7 2 8 2 1 6 6 2 1 3 7 6 3 4 2 7 5 1 4 4 1 7 3 8 3 5 5 4 8 8 4 1 2 1 2 7 6 2 2 5 5 2 3 8 3 5 4 6 3 4 8 5 5 4 6 2 4 4 4 7 6 1 2 3 7 3 8 2 1 4 6 5 5 1 7 1 7 7 7 1 8 5 2 3 6 8 1 3 3 4 7 3 6 5 6 6 2 2 8 6 4 1 5 7 7 6 1 6 4 4 8 3 3 2 8 3 5 5 4 4 7 4 5 1 3 3 3 2 7 2 6 6 7 5 1 7 8 2 7 3 6 8 6 7 6 4 5 5 5 2 8 1 4 4 4 4 7 7 2 8 4 2 3 5 3 2 5 1 7 1 3 7 7 6 4 5 8 1 1 8 4 2 2 5 3 2 3 3 6 1 1 6 5 6 6 6 3 4 2 4 2 8 8 1 1 5 3 2 3 1 6 6 6 3 2 7 5 5 2 3 5 8 1 7 5 4 4 5 4 4 4 2 1 3 3 8 8 3 6 6 7 1 5 5 6 7 2 6 6 8 7 6 6 7 2 2 5 6 3 2 1 6 4 6 6 6 3 2 5 3 5 7 5 4 5 7 5 2 5 7 7 6 6 7 4 1 7 5 5 4 4 4 4 8 7 3 3 3 1 7 7 8 8 8 2 1 3 3 2 2 8 2 2 4 2 7 5 7 6 2 7 1 1 7 8 8 4 5 5 6 3 1 2 6 4 3 7 3 6 3 4 4 7 1 1 5 1 6 3 4 5 3 8 3 7 8 8 8 2 2 7 7 8 6 8 6 5 7 4 2 1 1 4 2 2 4 4 4 1 8 3 3 5 5 6 6 5 6 6 5 8 2 1 6 6 6 6 3 3 1 5 5 3 6 6 2 3 3 4 5 3 3 1 7 7 4 4 7 3 2 1 5 5 5 1 6 7 2 4 6 2 5 5 6 1 7 4 1 1 6 7 5 2 3 2 7 7 7 4 4 7 4 6 5 2 2 1 1 4 4 4 3 2 6 6 5 7 5 8 8 3 3 3 8 6 5 6 5 6 4 4 4 5 7 6 4 2 3 3 3 7 4 1 1 1 7 1 1 5 7 7 6 2 1 7 7 1 1 1 1 8 8 8 4 8 8 5 4 2 8 8 5 1 4 4 7 7 7 1 2 2 8 8 2 2 6 3 3 2 2 2 5 2 2 5 2 2 2 4 7 7 3 3 5 6 6 6 5 2 3 3 8 8 8 8 3 3 3 7 6 5 5 7 1 3 8 8 3 4 4 4 4 5 5 8 8 7 3 3 5 7 4 4 6 5 6 6 6 3 4 2 4 4 7 6 6 3 2 4 4 1 1 2 2 6 6 1 1 2 2 1 1 5 5 1 1 8 8 2 2 4 7 7 7 3 3 3 6 4 4 8 8 7 7 7 1 1 5 5 7 7 3 3 2 7 7 2 2 2 3 3 5 5 6 6 1 1 1 4 1 6 6 3 5 4 6 4 1 4 4 4 7 7 7 1 3 3 1 7 7 1 6 6 8 1 1 6 5 6 1 6 5 5 6 6 6 5 5 6 3 5 8 8 8 8 3 3 7 7 5 5 8 8 1 1 5 5 5 7 4 3 3 3 6 5 6 4 4 2 2 2 5 3 5 2 2 2 4 4 8 7 7 4 2 2 6 3 2 7 4 6 6 3 6 6 4 3 6 4 5 5 3 5 4 6 5 6 6 6 1 1 1 3 7 1 6000 8000 10000 12000 m/z 9

10 12000 raw data--Detector Voltage 10000 m/z 8000 6000 0.05 0.04 0.03 0.02 0.01 0.0

Raw data: normalized intensity vs. m/z. 11 0.025 0.020 0.015 0.010 0.005 0.0 normalized intensity

Classical Polymer Approaches Classical Polymer Approaches M n = number-average molecular weight First moment – M w = weight-average molecular weight 2 nd moment/1 st moment – M z = z-average molecular weight Third moment/2 nd moment 5/28/04 Interface2004 12