Coordinating Hereditary and System Investigation to Describe Qualities Identified with Mouse Weight

0
0
1842 days ago, 635 views
PowerPoint PPT Presentation
http://www.genetics.ucla.edu/labs/horvath/CoexpressionNetwork ... Mouse hereditary qualities. Anatole Ghazalpour, Sud Doss, Bin Zhang, Chris Plaisier, Susanna Wang, Eric E Schadt (Merck) ...

Presentation Transcript

Slide 1

Coordinating Genetic and Network Analysis to Characterize Genes Related to Mouse Weight Steve Horvath University of California, Los Angeles

Slide 2

Contents Brief survey of quality system development New wording: Gene noteworthiness in light of body weight Module quantitative attribute locus (mQTL=eQTL hotspot for a given module) Gene hugeness measure in light of a SNP Characterize body weight related qualities in mice

Slide 3

Important Task in Many Genomic Applications: Given a system (pathway) of communicating qualities (proteins) how to locate the focal players?

Slide 4

Which of the accompanying mathematicians had the greatest impact on others? Availability can be an essential variable for distinguishing imperative hubs

Slide 5

Network Construction Bin Zhang and Steve Horvath (2005) "A General Framework for Weighted Gene Co-Expression Network Analysis", Statistical Applications in Genetics and Molecular Biology: Vol. 4: No. 1, Article 17.

Slide 6

Network=Adjacency Matrix A system can be spoken to by a contiguousness grid, A=[a ij ], that encodes whether/how a couple of hubs is associated. A will be a symmetric lattice with passages in [0,1] For unweighted organize, sections are 1 or 0 relying upon regardless of whether 2 hubs are nearby (associated) For weighted systems, the contiguousness grid reports the association quality between quality sets

Slide 7

Generalized Connectivity Gene availability = push aggregate of the nearness framework For unweighted networks=number of direct neighbors For weighted networks= entirety of association qualities to different hubs

Slide 8

Overview: quality co-expression arrange examination Steps for building a co-expression organize Microarray quality expression information Measure concordance of quality expression with a Pearson relationship C) The Pearson connection network is either dichotomized to land at a contiguousness grid ��  unweighted arrange Or changed ceaselessly with the power nearness work ��  weighted system

Slide 9

Power nearness work brings about a weighted quality system Often picking beta=6 functions admirably yet when all is said in done we utilize the "scale free topology paradigm" depicted in Zhang and Horvath 2005.

Slide 10

Comparing nearness capacities Power Adjancy versus Step Function

Slide 11

Comparing the power contiguousness capacity to the progression work While the system investigation results are typically very strong concerning the system development strategy there are a few purposes behind leaning toward the power contiguousness work. Exact discovering: Network results are profoundly hearty as for the decision of the power beta Theoretical discovering: Network Concepts bode well as far as the module eigengene.

Slide 12

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 Focus of this discussion: Relate the Network Concepts to External Gene or Sample Information

Slide 13

Integrating Genetic and Network Analysis to Characterize Genes Related to Mouse Weight A Ghazalpour, S Doss, B Zhang, C Plaisier, S Wang, EE Schadt, T Drake, AJ Lusis, S Horvath. PLoS Genetics August 2006

Slide 14

F2 mouse cross information We connected the system development calculation to a subset of quality expression information from a F2 intercross between ingrained strains C3H/HeJ and C57BL/6J. Utilized liver quality expression information from 135 female mice (altogether different from male mice!) Goal: Characterize qualities whose expression profile are corresponded with body weight Statistical Method: Integrate arrange ideas with hereditary ideas in a multivariate direct relapse display

Slide 15

Defining Gene Modules =sets of firmly co-managed qualities

Slide 16

Module Identification in view of the thought of topological cover One essential point of metabolic system examination is to identify subsets of hubs (modules) that are firmly associated with each other. We receive the meaning of Ravasz et al (2002): modules are gatherings of hubs that have high topological cover.

Slide 17

Topological Overlap prompts to a system remove measure Generalized in Zhang and Horvath (2005) to the instance of weighted systems Generalized in Yip and Horvath (2006) to higher request cooperations

Slide 18

Using the topological cover uniqueness network to bunch qualities To gathering hubs with high topological cover into modules (groups), we utilize normal linkage progressive grouping combined with the TOM difference measure. Modules relate to branches of the dendrogram Once a dendrogram is gotten from a various leveled bunching technique, modules compare to cut-off branches. we utilize the "dynamic tree cut calculation" since it takes into account an adaptable decision of stature shorts.

Slide 19

Module plots for female liver expression information

Slide 20

Mouse body weight offers ascend to a quality criticalness measure Abstract meaning of a quality centrality measure: GS(i) is non-negative, the greater, the more *biologically* noteworthy Example: GS(i)=-log(p-esteem) But here we utilize GSweight(i) = |cor(x(i), weight)| where x(i) is the quality expression profile of the ith quality.

Slide 21

A quality criticalness measure normally offers ascend to a module centrality measure Module Significance=mean quality importance

Slide 22

The blue module has high module hugeness as for body weight, i.e. it is exceptionally advanced with qualities that are corresponded with weight

Slide 23

Relating the blue module qualities to 22 physiological attributes

Slide 24

Message: unsupervised module location strategy found an organically intriguing module The system modules were characterized without respect to a physiological characteristic (unsupervised bunching of qualities) The blue module is involved qualities that identify with physiologically fascinating qualities, specifically body weight. Quality philosophy: The blue module is improved for qualities in the 'additional cell grid (ECM) receptor collaboration' (p=2.3x10-9) and 'supplement and coagulant falls' (p=1.0x10-6) pathways.

Slide 25

Since profoundly associated `hub' qualities have been observed to be organically essential in different applications, it is characteristic to ask whether GSweight is identified with intramodular availability in the blue module. Assist it is fascinating to think about the relationship amongst GSweight and k in various sexual orientation/tissue blends.

Slide 26

Relating blue module availability to weight-based quality hugeness in various sex/tissue blends. Message: there is a profoundly critical relationship amongst GSweight and k In the female liver system which can't be found in different blends.

Slide 27

Understanding the hereditary drivers of the module qualities Since hereditary marker information were accessible for every mouse, it is characteristic to relate blue module quality expressions to the SNP markers. This could distinguish the hereditary drivers of the blue module pathway. Utilizing 1065 single nucleotide polymorphism (SNP) markers that were equally divided over the genome (~1.5 cM thickness), we mapped the quality expression values and plotted the circulation of the expression quantitative characteristic loci (eQTL) for all qualities inside every quality module.

Slide 28

Comparing eQTL hotspots between the 3421 most associated qualities (dark) and the module qualities (blue)

Slide 29

Module QTLs=mQTL =chromosomal area that influences module quality expressions. we estimated that there might likewise be genomic problem areas which coordinately direct the transcript levels of the qualities inside every module. New Terminology: Module QTL (mQTL)=genomic "hotspot" that manages transcript levels of the module qualities.

Slide 30

Comparing the body weight LOD score bend (dark bend) to circulation of module eQTLs (blue bars) of the blue module Blue bar= No. of qualities whose expression LOD score at the marker >2 Red stars name mQTLs Message: While there is some cover between the mQTLs and clinical attributes (chromosome 19) there are additionally affirmed contrasts: see the blue spike (mQTL2) on chromosome 2.

Slide 31

A SNP marker actually offers ascend to a measure of quality importance Additive SNP marker coding: AA->2, AB->1, BB->0 Absolute estimation of the relationship guarantees this is equal to AA->0, AB->1, BB->2 Dominant or passive coding might be more proper in a few circumstances Conceptually identified with a LOD score at the SNP marker for the i-th quality expression characteristic GS.SNP(i) = |cor(x(i), SNP)|.

Slide 32

Using mQTLs to characterize quality criticalness measures GSmQTL2(i) = |cor(x(i), mQTL2)| GSmQTL5(i) = |cor(x(i), mQTL5)| GSmQTL10(i) = |cor(x(i), mQTL10)| GSmQTL19(i) = |cor(x(i), mQTL19)| We likewise think that its valuable to characterize the accompanying rundown covariate since it is very huge in our multivariate direct relapse demonstrate GSmQTL*(i)=GSmQTL2+GSmQTL5+GSmQTL10

Slide 33

Multivariate Linear Regression Models for GSweight

Slide 35

The coordinated model permits us to portray qualities that are identified with weight Here the blue module qualities are binned into 2^3=8 containers made by dichotomizing the covariates GSmQTL* (high=q+,low=q-), GSmQTL19(high19+), k(high=k+). (parts were picked by the middle)

Slide 36

Discussion The multivariate relapse models in the Table highlight the benefit of taking a system point of view. Demonstrate 3 coordinates co-expression organize ideas (network) and hereditary marker data (GSmQTL) to clarify 70% of the variety in GSweight. This basic model is alluring since it shows that 3 naturally instinctive factors suffice to clarify which qualities of this pathway are identified with body weight. Incorporating quality co-expression systems with hereditary marker data permits one to comprehend what variables impact the relationship between quality expression and weight.

Slide 37

Comparing our investigations to

SPONSORS