High Dimensional Biological Data Analysis and Visualization
•Download as PPTX, PDF•
3 likes•22,586 views
This document discusses metabolomic data analysis techniques for studying diseases. It analyzes over 13,000 biological samples per year using over 160,000 data points per study. Univariate and multivariate statistical analyses are described, with multivariate being preferred. Techniques include principal component analysis, partial least squares discriminant analysis, hierarchical clustering analysis, and pathway enrichment analysis. Visualization and network mapping tools are also discussed to identify relationships between altered metabolites and treatment effects.
Recommended
Recommended
More Related Content
What's hot
What's hot (20)
Similar to High Dimensional Biological Data Analysis and Visualization
Similar to High Dimensional Biological Data Analysis and Visualization (20)
More from Dmitry Grapov
More from Dmitry Grapov (8)
Recently uploaded
Recently uploaded (20)
High Dimensional Biological Data Analysis and Visualization
- 1. Metabolomic Data Analysis for the Study of Diseases Dmitry Grapov, PhD
- 2. State of the art facility producing massive amounts of biological data… >13,000 samples/yr >160 studies ~32,000 data points/study
- 3. Goals?
- 4. Analysis at the Metabolomic Scale
- 5. Univariate vs. Multivariate Multivariate Predictive Modeling Group 2 Group 1 Univariate Hypothesis testing (t-Test, ANOVA, etc.) PCA O-/PLS/-DA
- 7. Data Complexity Meta Data m n variables Experimental Design = complexity samples Data m-D 1-D 2-D Variable # = dimensionality
- 8. Statistical Analysis •Identify differences in sample population means •sensitive to distribution shape •parametric = assumes normality •error in Y, not in X (Y = mX + error) wide •optimal for long data •assumed independence •false discovery rate (FDR) long n-of-one
- 9. Achieving “significance” is a function of: significance level (α) and power (1-β ) effect size (standardized difference in means) sample size (n)
- 10. False Discovery Rate (FDR) Type I Error: False Positives •Type II Error: False Negatives •Type I risk = •1-(1-p.value)m m = number of variables tested FDR correction • p-value adjustment or estimate of FDR (Fdr, q-value) Bioinformatics (2008) 24 (12):1461-1462
- 11. FDR correction FDR adjusted p-value Benjamini & Hochberg (1995) (“BH”) •Accepted standard Bonferroni •Very conservative •adjusted p-value = p-value*# of tests (e.g. 0.005 * 148 = 0.74 ) p-value
- 12. Multivariate Analysis Clustering • Grouping based on similarity/dissimilarity Principal Components Analysis (PCA) • Identify modes of variance in the data Partial Least Squares (PLS) •Identify modes of variance in the data correlated with a hypothesis
- 13. Cluster Analysis Use similarity/dissimilarity to group a collection of samples or variables Linkage Approaches •hierarchical (HCA) •non-hierarchical (k-NN, k-means) •distribution (mixtures models) •density (DBSCAN) Distribution •self organizing maps (SOM) k-means Density
- 14. Hierarchical Cluster Analysis similarity/dissimilarity defines “nearness” or distance objects are grouped based on linkage methods
- 15. Hierarchy of Similarity How does my metadata match my data structure? Hierarchy of effect sizes x x x Similarity x
- 16. Projection of Data Raw data PCA dimensions http://www.scholarpedia.org/article/Eigenfaces The algorithm defines the position of the light source Principal Components Analysis (PCA) • unsupervised • maximize variance (X) Partial Least Squares Projection to Latent Structures (PLS) • supervised • maximize covariance (Y ~ X) PC1 PC2 James X. Li, 2009, VisuMap Tech.
- 17. Interpreting PCA Results Variance explained (eigenvalues) Row (sample) scores and column (variable) loadings
- 18. How are scores and loadings related?
- 19. Centering and Scaling PMID: 16762068
- 20. Use PLS to test a hypothesis Partial Least Squares (PLS) is used to identify planes of maximum correlation between X measurements and Y (hypothesis) PLS PCA time = 0 120 min.
- 21. PLS model validation is critical Determine in-sample (Q2) and outof-sample error (RMSEP) and compare to a random model •permutation tests •training/testing
- 22. Biochemical domain information Databases for organism specific biochemical information: Multiple organisms •KEGG •BioCyc •Reactome Human •HMDB •SMPDB
- 25. Data visualization as form of analysis Dextromethorphan = additives in DM Liver CYP2D6 •high fructose corn syrup dextrorphan • antioxidants •flavor
- 26. Identification of relationships between altered metabolites urea cycle protein glycosylation nucleotide synthesis
- 27. Identification of treatment effects
- 28. Analysis of differential metabolic responses Treatment 1 Treatment 2
- 29. Resources •DeviumWeb- Dynamic multivariate data analysis and visualization platform url: https://github.com/dgrapov/DeviumWeb •imDEV- Microsoft Excel add-in for multivariate analysis url: http://sourceforge.net/projects/imdev/ •MetaMapR: Network analysis tools for metabolomics url: https://github.com/dgrapov/MetaMapR •TeachingDemos- Tutorials and demonstrations •url: http://sourceforge.net/projects/teachingdemos/?source=directory •url: https://github.com/dgrapov/TeachingDemos •CDS Blog- Data analysis case studies url: http://imdevsoftware.wordpress.com/
- 30. dgrapov@ucdavis.edu metabolomics.ucdavis.edu This research was supported in part by NIH 1 U24 DK097154