Gene expression matrix annotation
So far we have considered gene expression matrices separately from gene and sample annotation. The use ofannotation has two applications. First, each gene or sample should be assigned a unique (standard) identifier, allowing re- Figure 4.8 An example of a gene expression graph or network, where nodes (circles) represent genes and edges (arrows connecting the circles) represent relationships between genes. searchers from different laboratories to communicate unambiguously (i.e. to make sure that...
Choice of array platform
The mediod of microarray manufacture and the nature of the substrate can be used to categorise arrays. First there is distinction between the so-called cDNA arrays, where polymerase chain reaction (PCR) products are typically used as sequences on the array features (spots), and oligonucleotide arrays, or oligo-arrays, where the features (spots) are made up of oligonucleotides. Oligo-arrays may be either spotted or synthesised in situ, i.e. the features can be made from presynthesised oligomers...
Data generation processing and analysis an overview
Every high throughput experiment consists of two major parts (i) material processing and data collection-, and (ii) information processing. In a microarray experiment material processing and data collection can be broken down into five steps 2 preparation of the biological samples to be studied 3 extraction and labelling of the RNA (or a representation of the RNA) from the samples 4 hybridisation of the labelled extracts to the array 5 scanning of the hybridised array. The scanned image is the...
A1 Statistical analysis BCLUST
A program to assess reliability of gene clusters from expression data by using a consensus tree and bootstrap resampling method, as described by Zhang and Zhao (2000). BRB ArrayTools is an integrated package, developed by Richard Simon and Amy Peng at the National Cancer Institute, for the visualisation and statistical analysis of DNA microarray gene expression data. The package uses an Excel front end with integrated analytical and visualisation tools.
Time series analysis
Time series experiments provide a particular type of gene expression profile, revealing information about the order and the time scale of the expression events. There are a number of ways one can treat time series diat would not be meaningful for odier types of expression profile. An obvious example of time series analysis is directed towards finding periodicity or a trend (Figure 4.18a). This approach was used by Spellman et al. (1998) for identifying genes whose expression correlates with...
Hierarchical agglomerative clustering
Hierarchical agglomerative clustering is a process in which the data are successively fused, typically until all the data points are included. For hierarchical agglomerative clustering usually all the pair-wise distances between objects need to be defined. An agglomerative process typically starts by considering each object data point as a separate, or singleton, cluster. Starting with n objects, the result of the first iteration of clustering is that the two objects that are most similar are...
Distance and similarity measures in expression space
Most of die gene expression data analysis methods are based on comparisons between die gene or sample expression profiles. In order to make these comparisons first we need a way to measure similarity or dissimilarity between these objects, i.e. between vectors representing genes or samples. Often it is easier to measure die distance between the objects (vectors in our case) instead of the similarity, though one can be transformed into the other. The distance between A and B, D(A, B), is said to...
Clusfavor
CLUSter and Factor Analysis using Varimax Ordiogonal Rotation performs cluster and factor analysis of gene expression data obtained from cDNA mi-croarrays Peterson, 2002 . The user can perform cluster analysis and varimax orthogonal rotation, view dendrograms, and run factor analysis on selected cluster-specific genes. An optional output contains matrices for the input data, distance matrices, factor loadings, eigen-values, eigen-vectors, and the percentage of total variation for genes within a...
What are microarrays and how do they work
A microarray is typically a glass or polymer slide, onto which DNA molecules are attached at fixed locations called spots or features in the context of microarrays these will be treated as synonyms . There may be tens of thousands of spots on an array, each containing tens of millions of identical DNA molecules or fragments of identical molecules , of lengths from tens to hundreds of nucleotides. For gene expression studies, each of these molecules should identify a single mRNA molecule, or...
Principal component analysis eigen vectors and eigengenes
Principal component analysis PCA is one of die most common methods used for gene expression data analysis, primarily to reduce the dimensionality of data and to find combinations of experiments or genes that joindy contribute most to variability in the data. Although the techniques used in PCA are not simple, the underlying idea is quite intuitive - it is based on finding the directions in multidimensional vector space that have the largest amplitude in the dispersion of data points. These...
