Overview

advertisement
Overview
Univariate - Multivariate Approaches:
Joint Modeling of Imaging & Genetic Data
•
•
•
•
•
•
•
•
•
•
•
Vince D. Calhoun, Ph.D.
Executive Science Officer &
Director, Image Analysis & MR Research
The Mind Research Network
Distinguished Professor, Electrical and Computer Engineering
(primary), Biology, Computer Science, Psychiatry, & Neurosciences
The University of New Mexico
Introduction
A Taxonomy of Approaches
Candidate gene analysis (c1)
A prior based multivariate analysis (on genetic, c2)
Data-driven multivariate analysis (on genetic, c2)
Component-based analysis (on imaging, c3)
Multivariate analysis across both (c4)
Multiple variables for brain + gene w/ priors (c4+c2)
Multivariate analysis across N>2 phenotypes (c4+)
Integration of additional bioinformatics resources
Challenges & Future Directions
1
2
Genetic Information
The First Challenge
• Genetic: single nucleotide polymorphism (SNP)
• Genetic: copy number variation (CNV)
deletion
insertion
• Epigenetic: methylation
3
4
Overview
•
•
•
•
•
•
•
•
•
•
•
5
Introduction
A Taxonomy of Approaches
Candidate gene analysis (c1)
A prior based multivariate analysis (on genetic, c2)
Data-driven multivariate analysis (on genetic, c2)
Component-based analysis (on imaging, c3)
Multivariate analysis across both (c4)
Multiple variables for brain + gene w/ priors (c4+c2)
Multivariate analysis across N>2 phenotypes (c4+)
Integration of additional bioinformatics resources
Challenges & Future Directions
6
1
A Taxonomy of Approaches
Overview
•
•
•
•
•
•
•
•
•
•
•
Introduction
A Taxonomy of Approaches
Candidate gene analysis (c1)
A prior based multivariate analysis (on genetic, c2)
Data-driven multivariate analysis (on genetic, c2)
Component-based analysis (on imaging, c3)
Multivariate analysis across both (c4)
Multiple variables for brain + gene w/ priors (c4+c2)
Multivariate analysis across N>2 phenotypes (c4+)
Integration of additional bioinformatics resources
Challenges & Future Directions
J. Liu and V. Calhoun, "A review of multivariate analyses in imaging genetics," Frontiers in Neuroinformatics,
vol. 8, pp. 1-11, 2014
7
8
Example: APOE4 in Alzheimer’s Disease
Candidate gene approach (c1)
9
Properties
10
Why Multivariate?
Cross–validation performance using the top M SNP’s selected
via two different methods as the basis for an N-factor principle
component model.
11
12
2
A prior based multivariate analysis (on genetic, c2)
Overview
Gene-set enrichment analysis
•
•
•
•
•
•
•
•
•
•
•
Introduction
A Taxonomy of Approaches
Candidate gene analysis (c1)
A prior based multivariate analysis (on genetic, c2)
Data-driven multivariate analysis (on genetic, c2)
Component-based analysis (on imaging, c3)
Multivariate analysis across both (c4)
Multiple variables for brain + gene w/ priors (c4+c2)
Multivariate analysis across N>2 phenotypes (c4+)
Integration of additional bioinformatics resources
Challenges & Future Directions
13
A. Subramanian, et al, Proc Natl Acad Sci U S A, vol. 102, pp. 15545-15550, Oct 25 2005
A prior based multivariate analysis (on genetic, c2)
14
Overview
• GSEA, Methylation, & Hippocampal Volume
•
•
•
•
•
•
•
•
•
•
•
Introduction
A Taxonomy of Approaches
Candidate gene analysis (c1)
A prior based multivariate analysis (on genetic, c2)
Data-driven multivariate analysis (on genetic, c2)
Component-based analysis (on imaging, c3)
Multivariate analysis across both (c4)
Multiple variables for brain + gene w/ priors (c4+c2)
Multivariate analysis across N>2 phenotypes (c4+)
Integration of additional bioinformatics resources
Challenges & Future Directions
Ehrlich, S, “Associations between DNA methylation and schizophrenia-related intermediate
functional and structural imaging phenotypes – a gene set enrichment analysis”, in preparation. 15
Data-driven multivariate analysis (on genetic, c2)
16
Methylation Gender Correction (c2)
• Reduction of genetic data
• Penalized regression (LASSO)
• PCA
• ICA
17
J. Liu, K. Hutchison, M. Morgan, N. I. Perrone-Bizzozero, J. Sui, and V. D. Calhoun, "Identification of
Genetic and Epigenetic Factors Contributing to Population Structure," PLoS ONE, vol. 5, pp. 1-8, 2010
18
3
Epigenetic-methylation study
•
•
Ancestry effect on methylation
Genomic ~27,000 sites from 23 pairs of chromosomes
130 subjects (heavy drinker 33 females, 97 males, age
31.39.7)
Goal: Association with gender, age, BMI, alcohol use, cigarette
use, marijuana use, depression, stress, etc.
Results:
Methylation correlation
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.95
0.55
0.49
0.4
Gender
Age*
BMI*
0.45
Max_drinks*
MJ_use*
*: after gender effect correction
J. Liu, K. Hutchison, M. Morgan, N. I. Perrone-Bizzozero, J. Sui, and V. D. Calhoun, "Identification of
Genetic and Epigenetic Factors Contributing to Population Structure," PLoS ONE, vol. 5, pp. 1-8, 2010
19
Introduction
A Taxonomy of Approaches
Candidate gene analysis (c1)
A prior based multivariate analysis (on genetic, c2)
Data-driven multivariate analysis (on genetic, c2)
Component-based analysis (on imaging, c3)
Multivariate analysis across both (c4)
Multiple variables for brain + gene w/ priors (c4+c2)
Multivariate analysis across N>2 phenotypes (c4+)
Integration of additional bioinformatics resources
Challenges & Future Directions
21
Overview
•
•
•
•
•
•
•
•
•
•
•
20
Component-based analysis (on imaging, c3)
Overview
•
•
•
•
•
•
•
•
•
•
•
J. Liu, K. Hutchison, M. Morgan, N. I. Perrone-Bizzozero, J. Sui, and V. D. Calhoun, "Identification of
Genetic and Epigenetic Factors Contributing to Population Structure," PLoS ONE, vol. 5, pp. 1-8, 2010
D. C. Glahn, A. M. Winkler, P. Kochunov, L. Almasy, R. Duggirala, M. A. Carless, J. C. Curran, R. L.
Olvera, A. R. Laird, S. M. Smith, C. F. Beckmann, P. T. Fox, and J. Blangero, "Genetic control over
the resting brain," Proc Natl Acad Sci U S A, vol. 107, pp. 1223-1228, Jan 19 2010
22
Multiple brain variables & multiple genetic variables (c4)
Introduction
A Taxonomy of Approaches
Candidate gene analysis (c1)
A prior based multivariate analysis (on genetic, c2)
Data-driven multivariate analysis (on genetic, c2)
Component-based analysis (on imaging, c3)
Multivariate analysis across both (c4)
Multiple variables for brain + gene w/ priors (c4+c2)
Multivariate analysis across N>2 phenotypes (c4+)
Integration of additional bioinformatics resources
Challenges & Future Directions
J. Liu and V. Calhoun, "A review of multivariate analyses in imaging genetics," Frontiers in Neuroinformatics,
vol. 8, pp. 1-11, 2014
23
24
4
RRR: one (of many) possible frameworks
fMRI/SNP connection
• fMRI component: specific brain regions with common
independent brain function.
• SNP component: a linearly weighted group of SNPs
functioning together.
• Relationship assumption
fMRI
GWAS
PPI database
Spatial ICA
Imaging Endo-phenotype
Extraction
If an association of SNPs partially define a certain brain function in
specific brain regions, Then, the linked fMRI and SNP components
should share a similar pattern of existence across subjects.
ICA Components
Genetic Network
construction
Genetic module extraction
Modules
fMRI components
Expression
pattern
SNP components
Subject 1
..
..
Subject N
Two sample t-test
Collaborative
sparse reduced
rank regression
Association analysis
Expression
pattern
Module,
ranking
Gene,
SNP
Functional
analysis
enrichment
25
Canonical Correlation Analysis (CCA) model
Parallel ICA: Two Goals
Data2: (SNP)
Data1: (fMRI)
X1
Identify
Hidden
Features
X2
W1
W2
A1
A2
S1
Identify
Hidden
Features
S2
Identify Linked Features
Illustration of combining both imaging and SNP data with the CCA
model to identify correlated genes and voxels.
MAX : {H(Y1) + H(Y2) },
<Infomax>
g()=Correlation(A1 ,A 2 ) 2 =
Subject to: arg max g{W1,W2 ŝ1,ŝ2 },
H. Cao, J. Duan, D. Lin, Y. Shugart, V. D. Calhoun, and Y.-P. Wang, "Sparse Representation Based Biomarker Selection for
Schizophrenia with Integrated Analysis of fMRI and SNPs," NeuroImage, in press,
Simulation
Cov(a1i ,a 2j ) 2
Var(a 1i )×Var(a 2j )
J. Liu, O. Demirci, and V. D. Calhoun, "A Parallel Independent Component Analysis Approach to Investigate
Genomic Influence on Brain Function," IEEE Signal Proc. Letters, vol. 15, pp. 413-416, 2008.
28
Initial Proof of Concept: SNP/fMRI Fusion
Simulation: Designed to provide a more complete understanding of Parallel ICA while applied to genomic SNP array
studies. We specified the parameters for each component and input them into PLINK, an open-source whole genome
association analysis toolset [http://pngu.mgh.harvard.edu/purcell/plink/].
Data Description: 20 Sz & 43 Healthy controls
fMRI: one image per subject (Target activation in AOD task)
SNP: one array per subject (384 SNP genotypes - -> 367 SNPs)
Conditions: sample size effect. case to control ratio, SNP array size effect, case-related SNP’s vs. total SNP’s, odds ratio,
connection strength between genotype and phenotype effects
Control vs Patient
p<0.001
Simulation results suggest that parallel ICA, in general, is able to extract more accurately the components and connections
than a correlation test, in particular for weak linkages. Results also indicate that the ratio of sample size to SNP size should
be at least 0.02. However, when the data have a low odds ratio or cases vs. controls ratio, the correlation test provides
results reliably, though with lower accuracy.
Liu, J, et al, IEEE Bioinformatics and Biomedicine. 2008: Philadelphia, PA.
29
SNP
Rs1466163
Rs2429511
Rs3087454
Rs821616
Rs885834
Rs1355920
R4765623
Rs4784642
Z score
-4.08
3.97
-3.09
2.96
-2.78
-2.77
2.73
-2.71
Rs2071521
Rs7520974
2.58
2.55
Gene
AADC: aromatic L-amino acid decarboxylase
ADRA2A: alpha-2A adrenergic receptor gene
CHRNA7: alpha 7 nicotinic cholinergic receptor
DISC1: disrupted in schizophrenia 1
CHAT: choline acetyltransferase
CHRNA7: cholinergic receptor, nicotinic, alpha 7
SCARB1: scavenger receptor class B, member 1
GNAO1: guanine nucleotide binding protein (G protein),
alpha activating activity polypeptide O
APOC3: apolipoprotein C-III
CHRM3: muscarinic-3 cholinergic receptor
J. Liu, G. D. Pearlson, A. Windemuth, G. Ruano, N. I. Perrone-Bizzozero, and V. D. Calhoun, "Combining
fMRI and SNP data to investigate connections between brain function and genetics using parallel ICA,"
Hum.Brain Map., vol. 30, pp. 241-255, 2009
30
5
SNP Selection and Approach
Number of Components
• Schizophrenia patients and healthy controls
• SNP data component number selection
• MCIC data: Boston, Iowa, Minnesota and New Mexico
• Genome-wide 1M SNP data - [biallelic coding (AA, AB, or BB]
• fMRI sensorimotor task- Block design motor response to auditory
stimulation
• 208 subjects with SNP (777365 SNPs) and fMRI data (52322 voxels)
• Component number estimated to be 8 for fMRI data (MDL)
• Increase component number from 2 to 60 for SNP data
• Compare the identified fMRI and SNP components obtained with
different SNP component numbers
• SNP selection
• Sliding window covering 5 consecutive component numbers
• Locate a region in which identified components remain stable
(select 5% most contributing SNPs from the identified component
and evaluate the overlapping ratios)
• SNPs differentiating schizophrenia patients and healthy controls are
included (p-value < 0.005, 3318 SNPs)
• SNPs related to schizophrenia risk genes are included (1843 SNPs
selected, related to DISC1, COMT, etc.)
• Combine selected SNPs: 5157 SNPs as final input
Reference SNP component number = 5
• Parallel ICA
SNP component number
Overlapping ratio
Correlation
6
0.843
0.982
7
0.927
0.984
8
0.851
0.957
9
0.970
0.951
• Use subject type (SZ patients. vs. healthy control) as reference
• Apply reference PCA to SNP and fMRI data to reduce dimension and
select components of interest
• Apply parallel ICA to identify linked components optimized to the
correlation between the two modalities
J. Chen, V. D. Calhoun, G. D. Pearlson, S. Ehrlich, J. Turner, B. C. Ho, T. Wassink, A. Michael, and J. Liu, "Multifaceted
genomic risk for brain function in schizophrenia," NeuroImage, vol. 61, pp. 866-875, 2012.
31
J. Chen, V. D. Calhoun, and J. Liu, "ICA Order Selection Based on Consistency: Application to Genotype Data," in Proc.
EMBS, San Diego, CA, 2012.
Consistency
Resulting Linked Component
• fMRI component number = 8, SNP component number = 5
• One pair of linked components is identified, with p-value passing
Bonferroni correction
• Leave-one-out validation
• Divide subjects into 10 sets
• Apply parallel-ICA to 90% of the data in each validation run
fMRI component SNP component
index
index
• Evaluate the consistency of the components based on correlations
between components identified in each validation run and the reference
components identified with the full dataset
1
2
……
32
Subjects
10
……
rfMRI-SNP
P-value
1
3
-0.065
3.49E-01
2
2
0.042
5.48E-01
3
1
0.099
1.51E-01
4
2
-0.138
4.54E-02
5
1
0.141
4.16E-02
6
5
-0.128
6.44E-02
7
1
0.178
9.95E-03
8
4
0.282
3.39E-05
• Bootstrap: Multiple runs of parallel-ICA with 5157 randomly selected
SNPs. The median correlation was 0.16
33
J. Chen, V. D. Calhoun, G. D. Pearlson, S. Ehrlich, J. Turner, B. C. Ho, T. Wassink, A. Michael, and J. Liu, "Multifaceted
genomic risk for brain function in schizophrenia," NeuroImage, vol. 61, pp. 866-875, 2012.
Validation Results
•
•
•
•
For each validation run, identify
the replicated fMRI/SNP
component exhibiting most similar
pattern to the reference fMRI/SNP
component identified with full
dataset
Calculate the fMRI-SNP
correlation between the replicated
components obtained in the
validation run
Determine if the replicated
components are the highest linked
pair (highest correlation)
A successful validation: both SNP
and fMRI components show
similar patterns to the reference
SNP and fMRI components,
respectively; and they are linked
34
Identified fMRI component
Index of
validation run
rSNP-fMRI
1
0.301
Highest linked
pair?
2
-0.295
√
3
-0.334
0.339
4
-0.301
√
5
0.246
√
6
-0.249
0.259
7
-0.306
√
8
0.347
√
Brain region
Brodmann area
Volume
Max z-score
9
0.387
√
Postcentral Gyrus
1, 2, 3, 5, 7, 40, 43
9.7/4.4
6.23(42,-32,60)/5.70(-39,-32,62)
10
0.304
√
Precentral Gyrus
4, 6
8.2/1.7
6.03(42,-12,56)/4.40(-59,-12,42)
Inferior Parietal Lobule
40
3.0/0.5
6.03(45,-35,57)/4.88(-45,-32,57)
Medial Frontal Gyrus
:6, 32
1.0/1.6
4.82(3,-3,53)/5.27(0,-3,53)
Superior Temporal Gyrus
21, 22, 38
1.4/0.5
3.24(62,-12,1)/2.93(-50,-3,-5)
√
J. Chen, V. D. Calhoun, G. D. Pearlson, S. Ehrlich, J. Turner, B. C. Ho, T. Wassink, A. Michael, and J. Liu, "Multifaceted
genomic risk for brain function in schizophrenia," NeuroImage, vol. 61, pp. 866-875, 2012.
35
J. Chen, V. D. Calhoun, G. D. Pearlson, S. Ehrlich, J. Turner, B. C. Ho, T. Wassink, A. Michael, and J. Liu, "Multifaceted
genomic risk for brain function in schizophrenia," NeuroImage, vol. 61, pp. 866-875, 2012.
36
6
Summary of SNP Results
ADNI Data
• Conduct pathway analysis and functional annotation clustering based
on identified 94 genes
• European-American ADNI subjects (N=757)
• IPA (Ingenuity Pathway Analysis) identifies “Schizophrenia of humans”
as one of the top biofunctions, involving 11 genes
• IPA also identifies a number of significant canonical pathways, four of
which are related to neurotransmitter signaling
• David’s Bioinformatics Resource reports the most significant cluster to be
functionally related to synapse. A cluster annotated as “cell projection” is
also identified
Disease and disorder
Gene
p-value
Schizophrenia of humans
BDNF, COMT, DISC1, DRD3, ERBB4, GAD1,
GRIN2B, HTR7, NOTCH4, NRG1, NRG3
6.49E-09
Neurotransmitter signaling pathway
Gene
p-value
GABA receptor signaling
Dopamine receptor signaling
GABRA4, GABRG3, GAD1
COMT, DRD3, PPP2R2C
2.13E-03
7.66E-03
Neuregulin signaling
Glutamate receptor signaling
NRG1, NRG3, ERBB4
GRIN2B, GRID2
1.15E-02
3.74E-02
Functional annotation cluster
Gene
p-value
Synapse
GABRA4, GABRG3, GAD1, GRIN2B, GRID2, ERBB4,
SHC4, OTOF, PSD3, CTBP2
DRD3, GAD1, GRIN2B, MYCBP2, DNAH11, WNT2,
ESR1, CDH13, ALCAM, MYO5A
5.20E-03
Cell projection
• 209 HC (Mean/SD Age = 76.05/4.94; 113 Males) with
no past history of neurological or psychiatric disorder,
• 367 subjects with MCI (Mean/SD Age = 74.95/7.37;
239 Males)
• 181 subjects (Mean/SD Age = 75.57/7.48; 100 Males)
with clinically-assessed AD
• SNP data
• 533,872 SNPs (after QA)
• sMRI data
• Baseline 1.5T MRI scans
• Segmented into GM maps
9.70E-02
J. Chen, V. D. Calhoun, G. D. Pearlson, S. Ehrlich, J. Turner, B. C. Ho, T. Wassink, A. Michael, and J. Liu, "Multifaceted
genomic risk for brain function in schizophrenia," NeuroImage, vol. 61, pp. 866-875, 2012.
37
ADNI Data
S. Meda, B. Narayanan, J. Liu, N. I. Perrone-Bizzozero, M. Stevens, V. D. Calhoun, D. C. Glahn, L. Shen, S. L.
Risacher, A. J. Sayking, and G. D. Pearlson, "A large scale multivariate parallel ICA method reveals novel imaginggenetic relationships for Alzheimer's Disease in the ADNI cohort," NeuroImage, vol. 60, pp. 1608-1621, 2012
S. Meda, B. Narayanan, J. Liu, N. I. Perrone-Bizzozero, M. Stevens, V. D. Calhoun, D. C. Glahn, L. Shen, S. L.
Risacher, A. J. Sayking, and G. D. Pearlson, "A large scale multivariate parallel ICA method reveals novel imaginggenetic relationships for Alzheimer's Disease in the ADNI cohort," NeuroImage, vol. 60, pp. 1608-1621, 2012
38
Genetic Components by Group
39
Alzheimer’s disease functional interaction pathway
S. Meda, B. Narayanan, J. Liu, N. I. Perrone-Bizzozero, M. Stevens, V. D. Calhoun, D. C. Glahn, L. Shen, S. L.
Risacher, A. J. Sayking, and G. D. Pearlson, "A large scale multivariate parallel ICA method reveals novel imaginggenetic relationships for Alzheimer's Disease in the ADNI cohort," NeuroImage, vol. 60, pp. 1608-1621, 2012
40
sMRI/SNP
Structural deficits in brain regions consistently implicated in previous schizophrenia reports, including frontal and
temporal lobes and thalamus were related to SNPs from 16 genes, several previously associated with schizophrenia
risk and/or involved in normal CNS development, including AKT, PI3k, SLC6A4, DRD2, CHRM2 and ADORA2A.
A. sMRI component –A (group difference)
B. sMRI component –B (linked, but no group difference)
modified from Sleegers K et al., TIGS 2009
41
K. Jagannathan, V. D. Calhoun, J. Gelernter, M. Stevens, J. Liu, F. Bolognani, A. Windemuth, G. Ruano, and G. D.
Pearlson, "Genetic associations of brain structural networks in schizophrenia: a preliminary study using parallel ICA,"
Biological Psychiatry, vol. 68, pp. 657-666, 2010
42
7
Genetics and P3 ERP generation
ERP Topography & SNP Associations
• Subjects: 41 healthy subjects (24 female, 17 male)
• EEG collected during AOD task,
target/novel ERPs extracted
Target Stimuli
SNPs
0.55
Novel Stimuli
SNPs
• Blood sample collected, genotyped 384 SNPs from 222
genes 6 physiological systems.
J. Liu, K. A. Kiehl, G. D. Pearlson, N. I. Perrone-Bizzozero, and V. D. Calhoun, "Genetic Determinants of
Target and Novelty Processing," NeuroImage, vol. 46, pp. 809-816, 2009.
Genes
rs1800545 ADRA2A
rs7412
APOE
rs1128503 ABCB1
rs6578993
TH
rs1045642 ABCB1
rs2278718 MDH1
rs4784642 GNAO1
rs521674 ADRA2A
Genes
rs1800545 ADRA2A
rs7412
APOE
rs6578993
TH
rs2278718 MDH1
rs1128503 ABCB1
rs429358
APOE
rs3813065 PIK3C3
rs4121817 PIK3C3
rs521674 ADRA2A
0.47
J. Liu, K. A. Kiehl, G. D. Pearlson, N. I. Perrone-Bizzozero, and V. D. Calhoun, "Genetic Determinants of
Target and Novelty Processing," NeuroImage, vol. 46, pp. 809-816, 2009.
43
Pathway Analysis
44
Classification with SNP & fMRI
SVM Classification
J. Liu, K. A. Kiehl, G. D. Pearlson, N. I. Perrone-Bizzozero, and V. D. Calhoun, "Genetic Determinants of
Target and Novelty Processing," NeuroImage, vol. 46, pp. 809-816, 2009.
Subjects
20 Patients
20 Healthy Controls
fMRI data
Auditory Oddball Task
Gene data
367 SNPs
Dataset
Number of
Training
Subjects
Number of
Testing
Subjects
1
4
36
2
12
28
3
20
20
4
28
12
5
36
4
H. Yang, J. Liu, J. Sui, G. Pearlson, and V. D. Calhoun, "A Hybrid Machine Learning Method for Fusing fMRI and
Genetic Data to Classify Schizophrenia," Frontiers in Human Neuroscience, vol. 4, pp. 1-9, 2010.
45
46
Overview
•
•
•
•
•
•
•
•
•
•
•
fMRI + Gene
fMRI
Gene
Constrained ICA
Introduction
A Taxonomy of Approaches
Candidate gene analysis (c1)
A prior based multivariate analysis (on genetic, c2)
Data-driven multivariate analysis (on genetic, c2)
Component-based analysis (on imaging, c3)
Multivariate analysis across both (c4)
Multiple variables for brain + gene w/ priors (c4+c2)
Multivariate analysis across N>2 phenotypes (c4+)
Integration of additional bioinformatics resources
Challenges & Future Directions
CDK14 (7q21)
ZNF879
GRM6 (5q35)
LOC 100506971
UGDH
SLC5
ZNF354C
ADAMTS2
47
J. Liu, M. Ghassemi, A. Michael, D. Boutte, W. Wells, N. I. Perrone-Bizzozero, F. Macciardi, D. Mathalon, J. Ford, S. Potkin,
J. Turner, FBIRN, and V. D. Calhoun, "An ICA with reference approach in identification of genetic variation and associated
48
brain networks," Frontiers in Human Neuroscience, vol. 6, pp. 1-10, 2012.
8
Parallel ICA w/ Reference (pICA-R)
pICA-R: Schizophrenia & ANK pathway
1
1/2  A 1/2
X 1/2  A 1/2  S 1/2 W

 S 1/2  W1/2  X 1/2
1
Y1/2 
, U1/2  W1/2 X  W10/20
1  e U1/2
F1  max  H Y1   max  E lnf y 1 Y1 
 



 

F2  max H Y2   dist 2 r, S 2k   max  E lnf y 2 Y2   λ r  (W2 X 2 )k
2

cov A 1i , A 2j  



2
F3  max  corr A 1i , A 2j    max 


 i, j

 i, j var A 1i var A 2j 

2
2

J. Chen, V. D. Calhoun, G. D. Pearlson, N. Perrone-Bizzozero, J. Sui, J. A. Turner, J. Bustillo, S. Ehrlich, S. Sponheim,
J. Canive, B. C. Ho, and J. Liu, "Guided Exploration of Genomic Risk for Gray Matter Abnormalities in Schizophrenia
Using Parallel Independent Component Analysis with Reference," NeuroImage, vol. 83, pp. 384-396, 2013.
49
J. Chen, V. D. Calhoun, G. D. Pearlson, N. Perrone-Bizzozero, J. Sui, J. A. Turner, J. Bustillo, S. Ehrlich, S. Sponheim,
J. Canive, B. C. Ho, and J. Liu, "Guided Exploration of Genomic Risk for Gray Matter Abnormalities in Schizophrenia
Using Parallel Independent Component Analysis with Reference," NeuroImage, in press
Overview
•
•
•
•
•
•
•
•
•
•
•
50
3-way parallel ICA
Introduction
A Taxonomy of Approaches
Candidate gene analysis (c1)
A prior based multivariate analysis (on genetic, c2)
Data-driven multivariate analysis (on genetic, c2)
Component-based analysis (on imaging, c3)
Multivariate analysis across both (c4)
Multiple variables for brain + gene w/ priors (c4+c2)
Multivariate analysis across N>2 phenotypes (c4+)
Integration of additional bioinformatics resources
Challenges & Future Directions
51
52
Results (N=112, fMRI, sMRI, 65K SNPs in coding genes)
(a)
sMRI Component
(b) fMRI Component
Corr = 0.44
Pval = 1.5e-5
Corr = 0.39
Pval = 9.0e-4
Genetics
Corr = 0.34
Pval = 2.0e-3
(c) Genetic Component
Functional
Activity
Structural
Variation
Overview
•
•
•
•
•
•
•
•
•
•
•
Introduction
A Taxonomy of Approaches
Candidate gene analysis (c1)
A prior based multivariate analysis (on genetic, c2)
Data-driven multivariate analysis (on genetic, c2)
Component-based analysis (on imaging, c3)
Multivariate analysis across both (c4)
Multiple variables for brain + gene w/ priors (c4+c2)
Multivariate analysis across N>2 phenotypes (c4+)
Integration of additional bioinformatics resources
Challenges & Future Directions
(d) Pairwise Connections
V. Vergara, A. Ulloa, V. D. Calhoun, D. Boutte, J. Chen, and J. Liu, "A Three-way Parallel ICA Approach to
Analyze Links among Genetics, Brain Structure and Brain Function," NeuroImage, in press,
54
9
55
Overview
•
•
•
•
•
•
•
•
•
•
•
56
Challenges & Issues
• Overfitting (cross-validation, prior information)
• Replication
• Interpretation (e.g. GSEA does not model exact
interaction among SNPs, latent component may
not have a biological purpose).
• Can be partially addressed by incorporating
known biological, cellular, or behavioral specific
information.
• Lots of information not fully incorporated
(proteomic, gene expression, epigenetic,
behavioral and environmental variables)
• CNV (low incidence)
Introduction
A Taxonomy of Approaches
Candidate gene analysis (c1)
A prior based multivariate analysis (on genetic, c2)
Data-driven multivariate analysis (on genetic, c2)
Component-based analysis (on imaging, c3)
Multivariate analysis across both (c4)
Multiple variables for brain + gene w/ priors (c4+c2)
Multivariate analysis across N>2 phenotypes (c4+)
Integration of additional bioinformatics resources
Challenges & Future Directions
57
58
Software
• http://mialab.mrn.org/software
• freeware, written in MATLAB
• Group ICA of fMRI Toolbox (GIFT)
•
•
•
•
•
•
•
Single subject/Group ICA
Many algorithms (e.g. ICA, cICA, IVA)
MANCOVA testing framework
Source Based Morphometry
Model order estimation
ICASSO (clustering/stability)
Dynamic connectivity (dFNC)
Left Hemisphere
Visual Stimuli Onset
• Fusion ICA Toolbox (FIT)
• Parallel ICA, jICA
• mCCA+jICA & much more!
• Simulation Toolbox (SimTB)
• Flexible generation of fMRI-like data
• Task & rest fMRI
• Space * Time * Amplitude
60
10
Download