STAT 254 -lecture1 An overview Cell biology, microarray, statistics Bioinformatics and Statistics Topics to cover Keep a skeptical eye on everything you read or hear Keep an eye on bigger picture; while working on specifics The shaping of bioinformatics falls on your shoulders What to take home : not just microarray, or high throughput data analysis methods, but a set of skills, ways of thinking about quantitative biology

Exploratory data analysis multivariate high dimensional 20 min

Study of Gene Expression: Statistics, Biology, and Microarrays Ker-Chau Li Statistics Department UCLA IMS ENAR Conference Time : March 31, 2003 Place:Tampa, FL

Outline Review of cell biology Microarray gene expression data collection Cell-cycle gene expression (Main Data set) PCA/Nested regression; SIR (Dim. red.) Similarity analysis - clustering (Why Popular?) Liquid association Closing remarks New statistical concept, fueled by Stein’s lemma Justification for IMS

PART I. Cellular Biology Macromolecules: DNA, mRNA, protein

Why Biology hot? Because of

Human Genome Project Begun in 1990, the U.S. Human Genome Project is a 13-year effort coordinated by the U.S. Department of Energy and the National Institutes of Health. The project originally was planned to last 15 years, but effective resource and technological advances have accelerated the expected completion date to Project goals are to Human Genome Program, U.S. Department of Energy, Genomics and Its Impact on Medicine and Society: A 2001 Primer, 2001 Recent Milestones: ■ June 2000 completion of a working draft of the entire human genome ■ February 2001 analyses of the working draft are published ■ identify all the approximate 30,000 genes in human DNA, ■ determine the sequences of the 3 billion chemical base pairs that make up human DNA, ■ store this information in databases, ■ improve tools for data analysis, ■ transfer related technologies to the private sector, and ■ address the ethical, legal, and social issues (ELSI) that may arise from the project.

Gene number, exact locations, and functions DNA sequence organization Chromosomal structure and organization Noncoding DNA types, amount, distribution, information content, and functions Interaction of proteins in complex molecular machines Evolutionary conservation among organisms Protein conservation (structure and function) Proteomes (total protein content and function) in organisms Correlation of SNPs (single-base DNA variations among individuals) with health and disease Disease-susceptibility prediction based on gene sequence variation Genes involved in complex traits and multigene diseases Complex systems biology including microbial consortia useful for environmental restoration Developmental genetics, genomics Future Challenges: What We Still Don’t Know Human Genome Program, U.S. Department of Energy, Genomics and Its Impact on Medicine and Society: A 2001 Primer, 2001 Predicted vs experimentally determined gene function {1} Gene regulation {2} (upstream regulatory region) Coordination of gene expression, protein synthesis, and post- translational events {3}

Medicine and the New Genomics Gene Testing Gene Therapy Human Genome Program, U.S. Department of Energy, Genomics and Its Impact on Medicine and Society: A 2001 Primer, 2001 improved diagnosis of disease earlier detection of genetic predispositions to disease rational drug design gene therapy and control systems for drugs Anticipated Benefits Pharmacogenomics personalized, custom drugs

Agriculture, Livestock Breeding, and Bioprocessing disease-, insect-, and drought-resistant crops healthier, more productive, disease-resistant farm animals more nutritious produce biopesticides edible vaccines incorporated into food products new environmental cleanup uses for plants like tobacco Human Genome Program, U.S. Department of Energy, Genomics and Its Impact on Medicine and Society: A 2001 Primer, 2001 Anticipated Benefits

How does the cell work? The guiding principle is the so-called

Medicine and the New Genomics Gene Testing Gene Therapy Human Genome Program, U.S. Department of Energy, Genomics and Its Impact on Medicine and Society: A 2001 Primer, 2001 improved diagnosis of disease earlier detection of genetic predispositions to disease rational drug design gene therapy and control systems for drugs Anticipated Benefits Pharmacogenomics personalized, custom drugs

Agriculture, Livestock Breeding, and Bioprocessing disease-, insect-, and drought-resistant crops healthier, more productive, disease-resistant farm animals more nutritious produce biopesticides edible vaccines incorporated into food products new environmental cleanup uses for plants like tobacco Human Genome Program, U.S. Department of Energy, Genomics and Its Impact on Medicine and Society: A 2001 Primer, 2001 Anticipated Benefits

How does the cell work? The guiding principle is the so-called

Human Genome Program, U.S. Department of Energy, Genomics and Its Impact on Medicine and Society: A 2001 Primer, 2001

Gene to protein 4 Nucleotides and 20 amino acids Protein is synthesized from amino acids by ribosome

Gene to Protein Transcription Translation

Transcription and translation

PART II. Microarray Genome-wide expression profiling

Exploring the Metabolic and Genetic Control of Gene Expression on a Genomic Scale Joseph L. DeRisi, Vishwanath R. Iyer, Patrick O. Brown*

Microarray

MicroArray Allows measuring the mRNA level of thousands of genes in one experiment -- system level response The data generation can be fully automated by robots Common experimental themes: –Time Course (when) –Tissue Type (where) –Response (under what conditions) –Perturbation: Mutation/Knockout, Knock-in Over-expression

Reverse-transcription Color : cy3, cy5 green, red

Example 1 Comparative expression Normal versus cancer cells ALL versus AML 5 min E.Lander’s group at MIT

PART III. Statistics Low-level analysis Comparative expression Feature extraction Clustering/classification Pearson correlation Liquid association

Issues related to image qualities Convert an image into a number representing the ratio of the levels of expression between red and green channels Color bias Spatial, tip, spot effects Background noises cDNA, oligonucleotide arrays, (not to be covered)

Genome-wide expression profile A basic structure cond1 cond2 …….. condp x11 x12 …….. x1p x21 x22 …….. x2p … …... xn1 xn2 …….. xnp Gene1 Gene2 Genen

Cond1, cond2, …, condp denote various environmental conditions, time points, cell types, etc. under which mRNA samples are taken Note : numerous cells are involved Data quality issues : 1. chip (manufacturer) 2. mRNA sample (user) It is important to have a homogeneous sample so that cellular signals can be amplified Yeast Cell Cycle data : ideally all cells are engaged in the same activities- synchronization

Two classes problem ALL (acute lymphoblastic leukemia) AML(acute myeloid leukemia) An application

Which Genes to select? For each gene (row) compute a score defined by sample mean of X - sample mean of Y divided by standard deviation of X + standard deviation of Y X=ALL, Y=AML Genes (rows) with highest scores are selected. Seems to work ! Improvement? 34 new leukemia samples 29 are predicated with 100% accuracy; 5 weak predication cases That seems to work well. They have a method

Study of cell-cycle regulated genes Rate of cell growth and division varies Yeast(120 min), insect egg(15-30 min); nerve cell(no);fibroblast(healing wounds) Regulation : irregular growth causes cancer Goal : find what genes are expressed at each state of cell cycle Yeast cells; Spellman et al (2000) Fourier analysis: cyclic pattern

Yeast Cell Cycle (adapted from Molecular Cell Biology, Darnell et al) Most visible event

Example of the time curve: Histone Genes: (HTT2) ORF: YNL031C Time course: Histone

EBP2: YKL172W TSM1: YCR042C YOR263C

Why clustering make sense biologically? Profile similarity implies functional association The rationale is Genes with high degree of expression similarity are likely to be functionally related and may participate in common pathways. They may be co-regulated by common upstream regulatory factors. Simply put, Rationale behind massive gene expression analysis:

Some protein complexes Protein rarely works as a single unit

Pearson's correlation coefficient, a simple way of describing the strength of linear association between a pair of random variables, has become the most popular measure of gene expression similarity. 1. Cluster analysis : average linkage, self-organizing map, K-mean, Classification : nearest neighbor,linear discriminant analysis, support vector machine,… 3. Dimension reduction methods : PCA ( SVD) Gene profiles and correlation

CC has been used by Gauss, Bravais, Edgeworth … Sweeping impact in data analysis is due to Galton( ) “ Typical laws of heridity in man ” Karl Pearson modifies and popularizes the use. A building block in multivariate analysis, of which clustering, classification, dim. reduct. are recurrent themes As a statistician, how can you ignore the time order ? (Isn’t it true that the use of sample correlation relies on the assumption that data are I.I.D. ???)

Other methods for Finding Gene clusters Bayesian clustering : normal mixture, (hidden) indicator PCA plot, projection pursuit, grand tour Multi-Dimension Scaling( bi-plot for categorical responses, showing both cases (genes) and variables(different clustering methods), displaying results from many different clustering procedures) Generalized association plot (Chen 2001, Statistica Sinica) PLAID model ( Statistica Sinica 2002, Lazzeroni, Owen)

1st PCA direction2nd PCA direction 3rd PCA direction Eigenvalues

Smooth S G1 S/G2 G2/M M/G S G1 S/G2 G2/M M/G Non-smooth Phase Assignment

ARG1 ARG2 Book a flight from LA to KEGG, JAPAN in less than 10 seconds Glutamate

ARG1 Adapted from KEGG X Y Compute LA(X,Y|Z) for all Z Rank and find leading genes 8th place negative

Coverage of bioinformatics by areas | topics Sequence analysis Microarray Linkage, pedigree DNA RNA Protein EST Drug Evolution Promoter 3-D structure Functional prediction Pathway discovery System modeling SNPAlternative splicing MotifDomain Drug -gene - protein Protein-protein TRANSFAC Protein -gene

Coverage of Bioinformatics by expertise (hat, not person) Biologist Computer scientist Statistician/m athematician (huge data volume) (raw data provider) Literature searching Make researcher’s life easier (pipeline) Data cleaning Data mining (Bio-information distilling/ Bio-data refining) Web page browsing Pattern searching /comparison Physical/Math/prob/stat models, computer optimization Gene Ontology Data base/ visualization Oil-refining(Crude oil) Generalization /inference (Noise, garbage, or ignorance?)

CurrentNext mRNA protein kinase Nutrients- carbon, nitrogen sources Temperature Water ATP, GTP, cAMP, etc localization DNA methylation, chromatin structure Math. Modeling : a nightmare FITNESSFITNESS FUNCTIONFUNCTION mRNA Cytoplasm Nucleus Mitochondria Vacuolar Observed hidden Statistical methods become useful

Bioinformatics (knowledge integration center) When Where Who What Why Cell level Organ level Organism level Species level Ecology system level

Special issue on bioinformatics Statistica Sinica 2002 January My paper on liquid association : PNAS 2002, 99, Want to get a quick start ? Genome-wide co-expression dynamics: theory and application Classification: Biological Science, Genetics; Physical Science, Statistics

END

Cautionary Notes for Seriation and row-column sorting Hierarchical clustering is popular, but Sharp boundaries may be artifacts due to “clever” permutation how to fine-tune user-specified parameters-need some theoretical guidance What is a cluster ? Criteria needed

Popular methods for clustering/data mining Linkage : Eisen et al, Alon et al K-mean : Tavazoein et al Self-organizing map : Tamayo et al SVD : Holter et al; Alter, Brown, Botstein

Can statisticians take the lead? Difficult But not impossible The key : Willingness to learn more biology February 2002, Talk at UCLA Biochemistry, feedback from David Eisenberg; March 2002, David gave an inspiring review talk about several of his works (Nature, similarity)