Microarray Analysis Jesse Mecham CS 601R

Microarray Analysis It all comes down to Experimental Design Experimental Design Preprocessing Preprocessing Data Analysis Data Analysis

Experimental Design Elimination of confounding factors Same cell line, minimal exposure Same cell line, minimal exposure Timing of sampling Timing of sampling Technological considerations Hybridization considerations Hybridization considerations Chip/tag selection Chip/tag selection

Slide to Data Gene Value D26528_at 193 D26561_cds1_at -70 D26561_cds2_at 144 D26561_cds3_at 33 D26579_at 318 D26598_at 1764 D26599_at 1537 D26600_at 1204 D28114_at 707

Preprocessing Data import Background adjustment Normalization Summarization of multiple probes per transcript Quality control

Data Import Incorporate various file formats into desired data formats Different vendors have different representations Different vendors have different representations Sometimes desired data is not provided Sometimes desired data is not provided

Background Adjustment It all comes down to one word…noise Optical distortion Optical distortion Non-specific hybridization Non-specific hybridization Equipment damage Equipment damage

M vs. A M represents differential ratio M = (log R – log G) A represents the fluorescence intensity A = (log R + log G)/2 Desirable transformation would show uniform distribution of differential across intensities

Normalization Normalization between samples needs to be established for a variety of reasons Different reverse transcription efficiency levels Different reverse transcription efficiency levels We are using PCR to amplify in separate plates Hybridization inequalities Hybridization inequalities Variations in solution used in hybridization reaction Spatial abnormalities between plates Spatial abnormalities between plates Particularly apparent for in-house plates

Background Example

Possible Problem in Background?

Summarizing Data Process of reducing the various samples into an analysis The crux of microarray analysis The crux of microarray analysis Can apply a linear or a non linear model using any of the following techniques Support Vector Machines (SVM) Support Vector Machines (SVM) Neural Networks Neural Networks Empirical Bayes Empirical Bayes

Quality Control Concerned with accuracy and reproducibility Dr. Piatetsy-Shapiro (last week’s colloquium) was primarily concerned with this area of microarray analysis Dr. Piatetsy-Shapiro (last week’s colloquium) was primarily concerned with this area of microarray analysis Detection of errors (x-validation) Isolation and validation of significant results Corrective behavior

Time for Fun Dataset ApoAI.RData ApoAI.RData The apolipoprotein AI (ApoAI) gene is known to play a pivotal role in high density lipoprotein (HDL) metabolism. Mice which have the ApoAI gene knocked (KO) out have very low HDL cholesterol levels. Puprose is to determine how ApoAI deficiency affects the action of other genes in the liver Help determine what molecular pathways ApoAI operates on Help determine what molecular pathways ApoAI operates on

Markers All mRNA data from both knockout and wild-type were marked GREEN KO and WT are marked RED Oftentimes, both populations are run on same plate with one being marked RED and the other marked GREEN Oftentimes, both populations are run on same plate with one being marked RED and the other marked GREEN

R “S”-like GNU project language and environment for statistical computing Great free package for linear and non- linear statistical modeling Also includes: an effective data handling and storage facility, an effective data handling and storage facility, a suite of operators for calculations on arrays, in particular matrices, a suite of operators for calculations on arrays, in particular matrices, a large, coherent, integrated collection of intermediate tools for data analysis, a large, coherent, integrated collection of intermediate tools for data analysis, graphical facilities for data analysis and display either on-screen or on hardcopy, and graphical facilities for data analysis and display either on-screen or on hardcopy, and a well-developed, simple and effective programming language which includes conditionals, loops, user-defined recursive functions and input and output facilities. a well-developed, simple and effective programming language which includes conditionals, loops, user-defined recursive functions and input and output facilities.

Bioconductor Open source package for statistical analysis of genomic data Includes both statistical and graphical tools Active project with a constant influx of new packages Does not include more complex analysis tools at this time (SVM’s, etc.)

With Controls

Controls Removed