Weighted gene co-expression analysis. Yuehua Cui June 7, 2013

Transcription

1 Weighted gene co-expression analysis Yuehua Cui June 7, 2013

2 Weighted gene co-expression network (WGCNA) A type of scale-free network: A scale-free network is a network whose degree distribution follows a power law, at least asymptotically. That is, the fraction P(k) of nodes in the network having k connections to other nodes goes for large values of k as p k = ck γ where c is a normalization constant and γ is a parameter whose value is typically in the range 2 < γ < 3, although occasionally it may lie outside these bounds. Scale-free networks are noteworthy because many empirically observed networks appear to be scale-free, including the world wide web, citation networks, biological networks, airline networks and some social networks. From Wikipedia

3 From Wikipedia

4 Reprinted from Linked: The New Science of Networks by Albert-Laszlo Barabasi Scale-Free Network Models in Epidemiology Adapted from J.B. Dunham and F.B. Berlin

5 Flight connections and hub airports In a scale-free network, the nodes with the largest number of links (connections) are most important! Courtesy of A. Barabasi

6 Note: The rest of the slides about WGCNA are adapted or modified from the slides in Dr. Steve Horvath s website at: xpressionnetwork/

7 Philosophy of Weighted Gene Co- Expression Network Analysis Understand the system instead of reporting a list of individual parts Describe the functioning of the engine instead of enumerating individual nuts and bolts Focus on modules as opposed to individual genes this greatly alleviates multiple testing problem Network terminology is intuitive to biologists

8 How to define a gene coexpression network?

9 Gene Co-expression Networks In gene co-expression networks, each gene corresponds to a node. Two genes are connected by an edge if their expression values are highly correlated. Definition of high correlation is somewhat tricky One can use statistical significance But we propose a criterion for picking threshold parameter: scale free topology criterion.

10 Frequency P(k) vs k in scale free networks P(k) Frequency Distribution of Connectivity Scale Free Topology refers to the frequency distribution of the connectivity k p(k)=proportion of nodes that have connectivity k Connectivity k

11 How to check Scale Free Topology? Idea: Log transformation p(k) and k and look at scatter plots Linear model fitting R 2 index can be used to quantify goodness of fit

12 Our `holistic view. Weighted Network View Unweighted View All genes are connected Connection Widths=Connection strengths Some genes are connected All connections are equal Hard thresholding may lead to an information loss. If two genes are correlated with r=0.79, they are deemed unconnected with regard to a hard threshold of τ=0.8

13 Network=Adjacency Matrix A network can be represented by an adjacency matrix, A=[a ij ], that encodes whether/how a pair of nodes is connected. A is a symmetric matrix with entries in [0,1] For unweighted network, entries are 1 or 0 depending on whether or not 2 nodes are adjacent (connected) For weighted networks, the adjacency matrix reports the connection strength between gene pairs

14 Connectivity Gene connectivity = row sum of the adjacency matrix For unweighted networks=number of direct neighbors For weighted networks= sum of connection strengths to other nodes Connectivity k a i i ij ji

15 How to construct a weighted gene co-expression network?

16 Using an adjacency function to define a network Measure co-expression by a similarity s(i,j) in [0,1] e.g. absolute value of the Pearson correlation Define an adjacency matrix as A(i,j) using an adjacency function AF(s(i,j)) Here we consider 2 classes of AFs Step function AF(s)=I(s>) with parameter (unweighted network) Power function AF(s)=s b with parameter b (weighted network) The choice of the AF parameters (, b) determines the properties of the network.

17 Power adjacency function results in a weighted gene network a cor( x, x ) ij i j Often choosing beta=6 works well but in general we use the scale free topology criterion described in Zhang and Horvath 2005.

18 Comparing the power adjacency functions with the step function Adjacency =connection strength Gene Co-expression Similarity

19 The scale free topology criterion for choosing the parameter values of an adjacency function. A) CONSIDER ONLY THOSE PARAMETER VALUES THAT RESULT IN APPROXIMATE SCALE FREE TOPOLOGY B) SELECT THE PARAMETERS THAT RESULT IN THE HIGHEST MEAN NUMBER OF CONNECTIONS Criterion A is motivated by the finding that most metabolic networks (including gene co-expression networks, proteinprotein interaction networks and cellular networks) have been found to exhibit a scale free topology Criterion B leads to high power for detecting modules (clusters of genes) and hub genes.

20 General Framework for Network Analysis Define a Gene Co-expression Similarity Define a Family of Adjacency Functions Determine the AF Parameters Define a Measure of Node Dissimilarity Identify Network Modules (Clustering) Relate Network Concepts to Each Other Relate the Network Concepts to External Gene or Sample Information

21 How to detect network modules?

22 Steps for defining gene modules Define a dissimilarity measure between the genes. Standard Choice: dissim(i,j)=1- abs(correlation) Choice by network community =1-Topological Overlap Matrix (TOM) Used here Use the dissimilarity in hierarchical clustering Define modules as branches of the hierarchical clustering tree Visualize the modules and the clustering results in a heatmap plot Heatmap

23 The topological overlap dissimilarity is used as input of hierarchical clustering TOM ij u a a a iu uj ij min( k, k ) 1a i j ij DistTOM ij 1TOM ij a cor( x, x ) ij i j Generalized in Zhang and Horvath (2005) to the case of weighted networks Generalized in Yip and Horvath (2006) to higher order interactions

24 Using the TOM matrix to cluster genes To group nodes with high topological overlap into modules (clusters), we typically use average linkage hierarchical clustering coupled with the TOM distance measure. Once a dendrogram is obtained from a hierarchical clustering method, we choose a height cutoff to arrive at a clustering. Here modules correspond to branches of the dendrogram Genes correspond to rows and columns TOM plot Hierarchical clustering dendrogram TOM matrix Module: Correspond to branches

25 Different Ways of Depicting Gene Modules Topological Overlap Plot Gene Functions 1) Rows and columns correspond to genes 2) Red boxes along diagonal are modules 3) Color bands=modules Multi Dimensional Scaling Traditional View Idea: Use network distance in MDS

26 Heatmap view of module Columns= tissue samples Rows=Genes Color band indicates module membership Message: characteristic vertical bands indicate tight co-expression of module genes

27 Module Eigengene= measure of over-expression=average redness=1 st PC of a given module Rows,=genes, Columns=microarray brown brown The brown module eigengenes across samples

28 Genes Gene expression database a conceptual view Samples Sample annotations Gene expression matrix Gene annotations Gene expression levels

29 Singular value decomposition (SVD) Use SVD to get the eigengenes Let X denote an m x n matrix of real-valued data and rank r m n, m genes and n samples The equation for singular value decomposition of X is the following: where U is an m x n matrix, S is an n x n diagonal matrix, and V T is also an n x n matrix. 29

30 UU T =V T V=I 30

31 T V U w n w X The w i are called the singular values of X If X is singular, some of the w i will be 0 In general rank(x) = number of nonzero w i SVD is mostly unique (up to permutation of singular values, or if some w i are equal) X -1 =(V T ) -1 S -1 U -1 = V S -1 U T Columns of V k corresponds to eigenvectors

32 Module eigengenes can be used to determine whether 2 modules are correlated. If correlation of MEs is high-> consider merging Martingale.Re ME.blue 0.08 ME.brow n ME.green Eigengenes can be used to build separate networks ME.grey ME.turquoise ME.yellow

33 Consensus eigengene networks in male and female mouse liver data and their relationship to physiological traits Langfelder P, Horvath S (2007) Eigengene networks for studying the relationships between co-expression modules. BMC Systems Biology 2007

34 Important Task in Many Genomic Applications: Given a network (pathway) of interacting genes how to find the central players? Gene connectivity = row sum of the adjacency matrix For unweighted networks=number of direct neighbors For weighted networks= sum of connection strengths to other nodes k i So value of k i indicates the important of the gene in a network j a ij

35 A Case Study MC Oldham, S Horvath, DH Geschwind (2006) Conservation and evolution of gene coexpression networks in human and chimpanzee brain. PNAS

36 What changed? Despite pronounced phenotypic differences, genomic similarity is ~96% (including single-base substitutions and indels) 1 Similarity is even higher in protein-coding regions 1 Cheng, Z. et al. Nature 437, (2005) Image courtesy of Todd Preuss (Yerkes National Primate Research Center)

37 Assessing the contribution of regulatory changes to human evolution Hypothesis: Changes in the regulation of gene expression were critical during recent human evolution (King & Wilson, 1975) Microarrays are ideally suited to test this hypothesis by comparing expression levels for thousands of genes simultaneously

38 Gene expression is more strongly preserved than gene connectivity Chimp Chimp Expression Cor=0.93 Cor=0.60 Human Expression Human Connectivity Hypothesis: molecular wiring makes us human Raw data from Khaitovich et al., 2004 Mike Oldham

39 A B Human Chimp

40 p = 1.33x10-4 p = 8.93x10-4 p = 1.35x10-6 p = 1.33x10-4

41 Connectivity diverges across brain regions whereas expression does not

42 Conclusions: chimp/human Gene expression is highly preserved across species brains Gene co-expression is less preserved Some modules are highly preserved Gene modules correspond roughly to brain architecture Species-specific hubs can be validated in silico using sequence comparisons

43 Software and Data Availability Sample data and R software tutorials can be found at the following webpage work An R package and accompanying tutorial can be found here: work/rpackages/wgcna/ Tutorial for this R package work/rpackages/wgcna/tutorialwgcnapackage.doc

44 What is different from other analyses? Emphasis on modules (pathways) instead of individual genes Greatly alleviates the problem of multiple comparisons Less than 20 comparisons versus comparisons Use of intramodular connectivity to find key drivers Quantifies module membership (centrality) Highly connected genes have an increased chance of validation Module definition is based on gene expression data No prior pathway information is used for module definition Two module (eigengenes) can be highly correlated Emphasis on a unified approach for relating variables Default: power of a correlation Rationale: puts different data sets on the same mathematical footing Considers effect size estimates (cor) and significance level p-values are highly affected by sample sizes (cor=0.01 is highly significant when dealing with observations) Technical Details: soft thresholding with the power adjacency function, topological overlap matrix to measure interconnectedness