Transcriptome Analysis Microarray Technology and Data Analysis Roy

Transcriptome Analysis Microarray Technology and Data Analysis Roy Williams Ph. D Sanford | Burnham Medical Research Institute

Microarray Revolution

Measuring Gene Expression Idea: measure the amount of m. RNA to see which genes are being expressed in (used by) the cell. Measuring protein would be more direct, but is currently harder.

General assumption of microarray technology �Use m. RNA transcript abundance level as a measure of expression for the corresponding gene �Proportional to degree of gene expression

How to measure RNA abundance �Several different approaches with similar themes �Illumina bead array – highly redundant oligo array �Affymetrix Gene. Chip – highly redundant oligo array �Nimblegen – highly redundant long oligo array � 2 -colour array (very long c. DNA; low redundancy) �SAGE (random Sanger sequencing of c. DNA library) �Reborn as Next Gen RNA seq

The Illumina Beadarray Technology �Highly redundant ~50 copies of a bead � 60 mer oligos �Absolute expression �Each array is deconvoluted using a colour coding tag system �Human, Mouse, Rat, Custom

Affymetrix Technology �Highly redundant (~25 short oligos per gene) �Absolute expression �PM-MM oligo system valuable for cross hybe detection �Human, Mouse, E. coli, Yeast……. . �Affy and illumina arrays have been systematically compared

Spotted Arrays �Low redundancy �c. DNA and oligo �Two dyes Cy 5/Cy 3 �Relative expression �Cost and custom

Single Colour Labelling

Microarrays in action off on

Areas Being Studied with Microarrays �Differential gene expression between two (or more) sample types �Similar gene expression across treatments �Tumour sub-class identification using gene expression profiles �Classification of malignancies into known classes �Identification of “marker” genes that characterize different cell types �Identification of genes associated with clinical outcomes (e. g. survival)

Experimental Design Experiment Perform Experiment • Replicates • 2 x 3 chips • <2 x 5 chips • Standardize conditions • Dump outliers

Microarray Data Analysis Workflow Quality Control Normalize Data Set up experiment al data Filter for differential expression Advanced analysis techniquesclustering Compare results to biology; Nextbio, Gene. Go; IPA

Recommended Software �Free Software – Gene. Pattern -- powerful, many plug-in packages and pipelines -- good video examples/tutorials �Gene. Spring GX 11 �R-Bioconductor (with guidance) �Hierarchical Cluster Explorer – easy clustering �Cytoscape, GSEA – for pathway visualisation �Partek �IPA, Nextbio, Gene. Go <= Burnham subscriptions!

Log Transformed Data 2/2 = 1 4/1=4 ¼=0. 25 log 2(1) = 0 log 2(4) = +2 log 2(0. 25) = -2 Transformation often performed before normalisation

BOXPLOT REPRESENTATION OF DATA SPREAD SIGNAL INTENSITY After QC for low confidence genes (P<0. 99) Note: ~50 replicate beads per array Outliers 75% quartile Median 25% quartile BAD CHIP NUMBER

The effect of quantiles Normalisation on the filtered 36 data sets IMPORTANT: use non-linear normalisation >library(affy) >Qdata <- normalize. quantiles(Rawdata) All same range

Data Analysis Examples � 1# Illumina arrays with Gene. Spring GX 11 � 2# Affymetrix data, with a Gene. Pattern module �Import, Quality Control, normalize �Detect differentially expressed genes �Pathway analysis

Illumina Analysis Workflow Genome Studio Application: process binary. idat files to txt Normalisation here is optional Check array hybridisation quality Direct Export file as “sample probe prof Import into GENESPRING GX 11

Gene. Spring GX 11 features �Guided workflows �Pathways �GSEA �IPA integration �Ontologies �My. SQL �R script API

Gene. Spring GX 11 �Create New Project �Browse to and load Data �Automated install of Genome. Def from Agilent repository

Illumina Advanced Workflow

Grouping Sample Replicates

Check Replicates Are Similar

Scatterplot of replicates

Scatterplot of differently treated samples

Filter genes on P-value

Significantly different genes in a Volcano plot

Significant Pathway Determination

Which types of genes are enriched in a cluster? �Idea: Compare your cluster of genes with lists of genes with common properties (function, expression, location). �Find how many genes overlap between your cluster and a gene list. �Calculate the probability of obtaining the overlap by chance This measures if the enrichment is significant. �This analysis provides an unbiased way of detecting connections between expression and function. Our Cell cycle 0 25 15000 7 Gene. Ontology Cell cycle

Send list to IPA for pathway Analysis

Significant Pathways sent to Ingenuity Pathway Analysis

Completed Analysis Data genelists Pathways

Affymetrix Workflow: Gene. Pattern

Comparative Marker Selection

Paste the URLs for Data files

Send results to next module Viewer module

Outputs ranked list of genes List of Marker genes can be Filtered and exported

Nextbio �Compares your Genelists to the Nextbio database �Can reveal unexpected similarities between datasets �Has a very good literature database connected to the results �Contains data from model organisms

Ingenuity Pathway Analysis �Detects networks in your data �Allows you to look for connections between genes and drugs/small molecules �Focused on Man and Mouse Gene. Go High Quality hand annotated ontologies �Has a very good literature database connected to the results �Contains data from model organisms

Start a new core analysis

Ingenuity Data import

IPA determines functions

Overlay drug and disease data

Data Import to Nextbio

The Nextbio Report Page

What else does my gene do?

THE END �Many thanks for coming!