Structured sparse CCA for brain imaging genetics via graph OSCAR

Lei Du, Heng Huang, Jingwen Yan, Sungeun Kim, Shannon Risacher, Mark Inlow, Jason Moore, Andrew Saykin, Li Shen

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

Background: Recently, structured sparse canonical correlation analysis (SCCA) has received increased attention in brain imaging genetics studies. It can identify bi-multivariate imaging genetic associations as well as select relevant features with desired structure information. These SCCA methods either use the fused lasso regularizer to induce the smoothness between ordered features, or use the signed pairwise difference which is dependent on the estimated sign of sample correlation. Besides, several other structured SCCA models use the group lasso or graph fused lasso to encourage group structure, but they require the structure/group information provided in advance which sometimes is not available. Results: We propose a new structured SCCA model, which employs the graph OSCAR (GOSCAR) regularizer to encourage those highly correlated features to have similar or equal canonical weights. Our GOSCAR based SCCA has two advantages: 1) It does not require to pre-define the sign of the sample correlation, and thus could reduce the estimation bias. 2) It could pull those highly correlated features together no matter whether they are positively or negatively correlated. We evaluate our method using both synthetic data and real data. Using the 191 ROI measurements of amyloid imaging data, and 58 genetic markers within the APOE gene, our method identifies a strong association between APOE SNP rs429358 and the amyloid burden measure in the frontal region. In addition, the estimated canonical weights present a clear pattern which is preferable for further investigation. Conclusions: Our proposed method shows better or comparable performance on the synthetic data in terms of the estimated correlations and canonical loadings. It has successfully identified an important association between an Alzheimer's disease risk SNP rs429358 and the amyloid burden measure in the frontal region.

Original languageEnglish (US)
Article number68
JournalBMC Systems Biology
Volume10
DOIs
StatePublished - Aug 26 2016

Fingerprint

Canonical Correlation Analysis
Neuroimaging
Brain
Imaging
Amyloid
Lasso
Imaging techniques
Group Structure
Graph in graph theory
Single Nucleotide Polymorphism
Synthetic Data
Weights and Measures
Genetic Association
Genes
Genetic Markers
Alzheimer's Disease
Information Structure
Alzheimer Disease
Signed
Pairwise

Keywords

  • Brain imaging genetics
  • Canonical correlation analysis
  • Machine learning
  • Structured sparse model

ASJC Scopus subject areas

  • Molecular Biology
  • Structural Biology
  • Applied Mathematics
  • Modeling and Simulation
  • Computer Science Applications

Cite this

Structured sparse CCA for brain imaging genetics via graph OSCAR. / Du, Lei; Huang, Heng; Yan, Jingwen; Kim, Sungeun; Risacher, Shannon; Inlow, Mark; Moore, Jason; Saykin, Andrew; Shen, Li.

In: BMC Systems Biology, Vol. 10, 68, 26.08.2016.

Research output: Contribution to journalArticle

Du, Lei ; Huang, Heng ; Yan, Jingwen ; Kim, Sungeun ; Risacher, Shannon ; Inlow, Mark ; Moore, Jason ; Saykin, Andrew ; Shen, Li. / Structured sparse CCA for brain imaging genetics via graph OSCAR. In: BMC Systems Biology. 2016 ; Vol. 10.
@article{0ef313d43691405bb26214510c027da3,
title = "Structured sparse CCA for brain imaging genetics via graph OSCAR",
abstract = "Background: Recently, structured sparse canonical correlation analysis (SCCA) has received increased attention in brain imaging genetics studies. It can identify bi-multivariate imaging genetic associations as well as select relevant features with desired structure information. These SCCA methods either use the fused lasso regularizer to induce the smoothness between ordered features, or use the signed pairwise difference which is dependent on the estimated sign of sample correlation. Besides, several other structured SCCA models use the group lasso or graph fused lasso to encourage group structure, but they require the structure/group information provided in advance which sometimes is not available. Results: We propose a new structured SCCA model, which employs the graph OSCAR (GOSCAR) regularizer to encourage those highly correlated features to have similar or equal canonical weights. Our GOSCAR based SCCA has two advantages: 1) It does not require to pre-define the sign of the sample correlation, and thus could reduce the estimation bias. 2) It could pull those highly correlated features together no matter whether they are positively or negatively correlated. We evaluate our method using both synthetic data and real data. Using the 191 ROI measurements of amyloid imaging data, and 58 genetic markers within the APOE gene, our method identifies a strong association between APOE SNP rs429358 and the amyloid burden measure in the frontal region. In addition, the estimated canonical weights present a clear pattern which is preferable for further investigation. Conclusions: Our proposed method shows better or comparable performance on the synthetic data in terms of the estimated correlations and canonical loadings. It has successfully identified an important association between an Alzheimer's disease risk SNP rs429358 and the amyloid burden measure in the frontal region.",
keywords = "Brain imaging genetics, Canonical correlation analysis, Machine learning, Structured sparse model",
author = "Lei Du and Heng Huang and Jingwen Yan and Sungeun Kim and Shannon Risacher and Mark Inlow and Jason Moore and Andrew Saykin and Li Shen",
year = "2016",
month = "8",
day = "26",
doi = "10.1186/s12918-016-0312-1",
language = "English (US)",
volume = "10",
journal = "BMC Systems Biology",
issn = "1752-0509",
publisher = "BioMed Central",

}

TY - JOUR

T1 - Structured sparse CCA for brain imaging genetics via graph OSCAR

AU - Du, Lei

AU - Huang, Heng

AU - Yan, Jingwen

AU - Kim, Sungeun

AU - Risacher, Shannon

AU - Inlow, Mark

AU - Moore, Jason

AU - Saykin, Andrew

AU - Shen, Li

PY - 2016/8/26

Y1 - 2016/8/26

N2 - Background: Recently, structured sparse canonical correlation analysis (SCCA) has received increased attention in brain imaging genetics studies. It can identify bi-multivariate imaging genetic associations as well as select relevant features with desired structure information. These SCCA methods either use the fused lasso regularizer to induce the smoothness between ordered features, or use the signed pairwise difference which is dependent on the estimated sign of sample correlation. Besides, several other structured SCCA models use the group lasso or graph fused lasso to encourage group structure, but they require the structure/group information provided in advance which sometimes is not available. Results: We propose a new structured SCCA model, which employs the graph OSCAR (GOSCAR) regularizer to encourage those highly correlated features to have similar or equal canonical weights. Our GOSCAR based SCCA has two advantages: 1) It does not require to pre-define the sign of the sample correlation, and thus could reduce the estimation bias. 2) It could pull those highly correlated features together no matter whether they are positively or negatively correlated. We evaluate our method using both synthetic data and real data. Using the 191 ROI measurements of amyloid imaging data, and 58 genetic markers within the APOE gene, our method identifies a strong association between APOE SNP rs429358 and the amyloid burden measure in the frontal region. In addition, the estimated canonical weights present a clear pattern which is preferable for further investigation. Conclusions: Our proposed method shows better or comparable performance on the synthetic data in terms of the estimated correlations and canonical loadings. It has successfully identified an important association between an Alzheimer's disease risk SNP rs429358 and the amyloid burden measure in the frontal region.

AB - Background: Recently, structured sparse canonical correlation analysis (SCCA) has received increased attention in brain imaging genetics studies. It can identify bi-multivariate imaging genetic associations as well as select relevant features with desired structure information. These SCCA methods either use the fused lasso regularizer to induce the smoothness between ordered features, or use the signed pairwise difference which is dependent on the estimated sign of sample correlation. Besides, several other structured SCCA models use the group lasso or graph fused lasso to encourage group structure, but they require the structure/group information provided in advance which sometimes is not available. Results: We propose a new structured SCCA model, which employs the graph OSCAR (GOSCAR) regularizer to encourage those highly correlated features to have similar or equal canonical weights. Our GOSCAR based SCCA has two advantages: 1) It does not require to pre-define the sign of the sample correlation, and thus could reduce the estimation bias. 2) It could pull those highly correlated features together no matter whether they are positively or negatively correlated. We evaluate our method using both synthetic data and real data. Using the 191 ROI measurements of amyloid imaging data, and 58 genetic markers within the APOE gene, our method identifies a strong association between APOE SNP rs429358 and the amyloid burden measure in the frontal region. In addition, the estimated canonical weights present a clear pattern which is preferable for further investigation. Conclusions: Our proposed method shows better or comparable performance on the synthetic data in terms of the estimated correlations and canonical loadings. It has successfully identified an important association between an Alzheimer's disease risk SNP rs429358 and the amyloid burden measure in the frontal region.

KW - Brain imaging genetics

KW - Canonical correlation analysis

KW - Machine learning

KW - Structured sparse model

UR - http://www.scopus.com/inward/record.url?scp=84983607522&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84983607522&partnerID=8YFLogxK

U2 - 10.1186/s12918-016-0312-1

DO - 10.1186/s12918-016-0312-1

M3 - Article

VL - 10

JO - BMC Systems Biology

JF - BMC Systems Biology

SN - 1752-0509

M1 - 68

ER -