Is Forkhead Box N1 (FOXN1) significant in both men and women
diagnosed with Chronic Fatigue Syndrome?
Charlyn Suarez
Mentors
Jeanette Papp
Anja Presson
Why?
Males and females are genetically different
Studies have shown genetic factors associated with a disease and/or a drug response appear to be different between male and female patients
Outline
Define Chronic Fatigue Syndrome
Dataset/Analyses
Weighted Gene Co-Expression Network Method
Previous Analysis
My Analysis
Conclusion
Additional Analyses
Future Goal
Chronic Fatigue Syndrome(CFS)
Complex disease characterized by profound fatigue Not improved by bed rest Worsened by physical and mental activities
Affects women at four times the rate of men
Cause remains unknown Genetics and environment The roles of the immune, endocrine and nervous
systems
Source: http://www.cdc.gov/cfs
Dataset
Comes from the CDC Chronic Fatigue Syndrome Research Group Contains microarray, SNP, and clinical data
Consists of 98 women and 29 men
Is restricted to genes that showed some sign of differential expression between CFS patients and controls
Previous Analysis
Has shown that FOXN1 is a candidate gene for CFS using the Weighted Gene Co-Expression Network Method (Presson et al., CAMDA 2006)
FOXN1 is differentially expressed in CFS patients and controls
R
Biological Significance of FOXN1
Mutations in mice & humans cause: Nudity Depleted immune system due to dysfunctional T-cells
Highly expressed in thymus epithelia cells: Convert lymphocytes to T-cells Release functional T-cells to fight infection
(Nehls et al. 1994; Pignata et al., 1996; Adriani et al. 2004)
CFS patients have an overactive immune system & high T-cell production (Maher et al. 2005)
Analysis
determine if FOXN1 is significant if the dataset is restricted to men or women
form a Weighted Gene Co-Expression Network for men and women separately
R
Overview of Weighted Gene Co-Expression Network
Developed by Steve HorvathBiostatistics & Human Genetics Department
University of California, Los Angeles
Biology and Networks
components of a living cell are dynamically interconnectedencoded into a complex intracellular web of molecular interactioncan be represented as a networkconnectivity within networks can be an important variable for identifying important nodes (genes)
Gene Co-Expression Network
each gene corresponds to a node
two genes are connected by an edge if their expression values are correlated
Networks can be represented by an adjacency matrix, A = [aij]
aij=connection strength between a pair of genes (0 ≤ aij ≤ 1 for all 1 ≤ i,j ≤ n)
V1 V2 V3 V4 V5
V1 1 .2 0 .1 .5
V2 0 .2 0 .3 0
V3 0 0 .8 0 .1
V4 0 0 0 1 0
V5 0 .9 0 0 1
Connection Strength
Gene
X
Gene
Y
Sample 1 1 2
Sample 2 2 5
Sample 3 3 6
Gene X Y
X 1 .9608
Y .9608 1
|cor(x,y)|
|cor(x,y)| 14
Gene X Y
X 1 .5713
Y .5713 1
Identifying Modules
Gene co-expression modules in the network were identified using average linkage hierarchical clustering
Modules for Original Analysis
Branches-clusters of similarly expressed genes
Modules-branches of the dendrogram
Trimmed to define 4 modules
grey color-genes that did not belong to any module
dendrogram
Gene Significance
0.00
0.05
0.10
0.15
0.20
CLUSTER0 , p-value= 2.5e-63
modules
correlation between gene expression and cluster trait (severity of CFS)
mea
n ge
ne s
igni
fica
nce
Connectivity/Gene Selection Criteria
Intramodular connectivity for each gene is the sum of the connection strengths between that gene and all other genes in its moduleFOXN1 is fairly connected in green moduleOther selection criteria in original gene selection Trait correlation SNP correlation(Presson et al, CAMDA 2006)
Modules for Males(dendrogram)
Gene Significance for Males
blue brown grey turquoise
0.00
0.05
0.10
0.15
0.20
CLUSTER0 - Males , p-value= 3.4e-33
modules
mea
n ge
ne
sign
ific
ance
Modules for Females
dendrogram
modules
(dendrogram)
Gene Significance for Females0.00
0.05
0.10
0.15
0.20
CLUSTER0 - Females , p-value= 2.6e-74
modules
mea
n ge
ne s
igni
fica
nce
Conclusions
FOXN1 is associated with CFS severity in men and women
FOXN1 is differentially expressed in men and women (higher expression in women)
Additional Analysis
Compare the gene significance between men and women in the original green module
Determine if the module structure from the combined analysis is preserved in the male and female analyses
Future Goal
To include my analysis in a paper that will be published in the proceedings for the Critical Assessment of Microarray Data Analysis (CAMDA)
References
http://www.cdc.gov/cfs
Anja Presson, Eric Sobel, Jeanette Papp, Aldons J. Lusis, Steve Horvath. Integration of Genetic and Genomic Approaches for the Analysis of Chronic Fatigue Syndrome Implicates Forkhead Box N1. http://www.camda.duke.edu/camda06/papers/
Pinsonneault J, Sadée W. Pharmacogenomics of Multigenic Diseases: Sex-Specific Differences in Disease and Treatment Outcome. AAPS PharmSci. 2003; 5 (4): article 29. DOI: 10.1208/ps050429
http://www.blackwellpublishing.com/press/pressitem.asp?ref=832 (Higher Mortality Rate For Females Undergoing Heart Surgery)
Acknowledgement
UCLA Genotyping &Sequencing Core Anja Presson Jeanette Papp Steve Horvath
SoCalBSI Jamil Momand Wendie Johnston Sandra Sharp Nancy Warter- Perez
NIH/NSF
Connection Strength
absolute value of the Pearson correlation coefficient was calculated for all pair-wise comparisons of gene-expression values across all microarray samples
correlation matrix was then transformed into a matrix of connection strengths using a power function (aij = |cor(xi, xj)|
β)
Dendrogram
Hierarchical clustering may be represented by a two dimensional diagram known as dendrogram
A dendrogram is a tree diagram frequently used to illustrate the arrangement of the clusters produced by a clustering algorithm (see cluster analysis)
Dendrograms are often used in computational biology to illustrate the clustering of genes
Hierarchical Clustering
Central to all of the goals of cluster analysis is the notion of degree of similarity (or dissimilarity) between the individual objects being clustered
agglomerative methods-proceed by series of fusions of the n objects into groups
divisive methods-separate n objects successively into finer groupings
Hierarchical Clustering