Wang Cong1, Xianglian Meng1,2, Jin Li1, Qiushi Zhang1,3, Feng Chen1, Wenjie Liu1, Ying Wang1, Sipu Cheng1, Xiaohui Yao4,5, Jingwen Yan4,5,6, Sungeun Kim4,6, Andrew J Saykin4,6, Hong Liang7, Li Shen8,9. 1. College of Automation, Harbin Engineering University, 145 Nantong Street, BLDG 61-5029, Harbin, 150001, China. 2. Harbin Huade University, No.288 Xue Yuan Rd. Limin Development Zone, Harbin, 150025, China. 3. College of Information Engineering, Northeast Dianli University, 169 Changchun Street, Jilin City, Jilin, 132012, China. 4. Department of Radiology and Imaging Sciences, Indiana University School of Medicine, 355 W 16th St, Suite 4100, Indianapolis, IN, 46202, USA. 5. School of Informatics and Computing, Indiana University, 719 Indiana Avenue, Indianapolis, IN, 46202, USA. 6. Indiana University Network Science Institute, Bloomington, IN, 47405, USA. 7. College of Automation, Harbin Engineering University, 145 Nantong Street, BLDG 61-5029, Harbin, 150001, China. lh@hrbeu.edu.cn. 8. Department of Radiology and Imaging Sciences, Indiana University School of Medicine, 355 W 16th St, Suite 4100, Indianapolis, IN, 46202, USA. shenli@iu.edu. 9. School of Informatics and Computing, Indiana University, 719 Indiana Avenue, Indianapolis, IN, 46202, USA. shenli@iu.edu.
Abstract
BACKGROUND: The cerebrospinal fluid (CSF) levels of total tau (t-tau) and Aβ1-42 are potential early diagnostic markers for probable Alzheimer's disease (AD). The influence of genetic variation on these CSF biomarkers has been investigated in candidate or genome-wide association studies (GWAS). However, the investigation of statistically modest associations in GWAS in the context of biological networks is still an under-explored topic in AD studies. The main objective of this study is to gain further biological insights via the integration of statistical gene associations in AD with physical protein interaction networks. RESULTS: The CSF and genotyping data of 843 study subjects (199 CN, 85 SMC, 239 EMCI, 207 LMCI, 113 AD) from the Alzheimer's Disease Neuroimaging Initiative (ADNI) were analyzed. PLINK was used to perform GWAS on the t-tau/Aβ1-42 ratio using quality controlled genotype data, including 563,980 single nucleotide polymorphisms (SNPs), with age, sex and diagnosis as covariates. Gene-level p-values were obtained by VEGAS2. Genes with p-value ≤ 0.05 were mapped on to a protein-protein interaction (PPI) network (9,617 nodes, 39,240 edges, from the HPRD Database). We integrated a consensus model strategy into the iPINBPA network analysis framework, and named it as CM-iPINBPA. Four consensus modules (CMs) were discovered by CM-iPINBPA, and were functionally annotated using the pathway analysis tool Enrichr. The intersection of four CMs forms a common subnetwork of 29 genes, including those related to tau phosphorylation (GSK3B, SUMO1, AKAP5, CALM1 and DLG4), amyloid beta production (CASP8, PIK3R1, PPA1, PARP1, CSNK2A1, NGFR, and RHOA), and AD (BCL3, CFLAR, SMAD1, and HIF1A). CONCLUSIONS: This study coupled a consensus module (CM) strategy with the iPINBPA network analysis framework, and applied it to the GWAS of CSF t-tau/Aβ1-42 ratio in an AD study. The genome-wide network analysis yielded 4 enriched CMs that share not only genes related to tau phosphorylation or amyloid beta production but also multiple genes enriching several KEGG pathways such as Alzheimer's disease, colorectal cancer, gliomas, renal cell carcinoma, Huntington's disease, and others. This study demonstrated that integration of gene-level associations with CMs could yield statistically significant findings to offer valuable biological insights (e.g., functional interaction among the protein products of these genes) and suggest high confidence candidates for subsequent analyses.
BACKGROUND: The cerebrospinal fluid (CSF) levels of total tau (t-tau) and Aβ1-42 are potential early diagnostic markers for probable Alzheimer's disease (AD). The influence of genetic variation on these CSF biomarkers has been investigated in candidate or genome-wide association studies (GWAS). However, the investigation of statistically modest associations in GWAS in the context of biological networks is still an under-explored topic in AD studies. The main objective of this study is to gain further biological insights via the integration of statistical gene associations in AD with physical protein interaction networks. RESULTS: The CSF and genotyping data of 843 study subjects (199 CN, 85 SMC, 239 EMCI, 207 LMCI, 113 AD) from the Alzheimer's Disease Neuroimaging Initiative (ADNI) were analyzed. PLINK was used to perform GWAS on the t-tau/Aβ1-42 ratio using quality controlled genotype data, including 563,980 single nucleotide polymorphisms (SNPs), with age, sex and diagnosis as covariates. Gene-level p-values were obtained by VEGAS2. Genes with p-value ≤ 0.05 were mapped on to a protein-protein interaction (PPI) network (9,617 nodes, 39,240 edges, from the HPRD Database). We integrated a consensus model strategy into the iPINBPA network analysis framework, and named it as CM-iPINBPA. Four consensus modules (CMs) were discovered by CM-iPINBPA, and were functionally annotated using the pathway analysis tool Enrichr. The intersection of four CMs forms a common subnetwork of 29 genes, including those related to tau phosphorylation (GSK3B, SUMO1, AKAP5, CALM1 and DLG4), amyloid beta production (CASP8, PIK3R1, PPA1, PARP1, CSNK2A1, NGFR, and RHOA), and AD (BCL3, CFLAR, SMAD1, and HIF1A). CONCLUSIONS: This study coupled a consensus module (CM) strategy with the iPINBPA network analysis framework, and applied it to the GWAS of CSF t-tau/Aβ1-42 ratio in an AD study. The genome-wide network analysis yielded 4 enriched CMs that share not only genes related to tau phosphorylation or amyloid beta production but also multiple genes enriching several KEGG pathways such as Alzheimer's disease, colorectal cancer, gliomas, renal cell carcinoma, Huntington's disease, and others. This study demonstrated that integration of gene-level associations with CMs could yield statistically significant findings to offer valuable biological insights (e.g., functional interaction among the protein products of these genes) and suggest high confidence candidates for subsequent analyses.
Authors: N Joan Abbott; Adjanie A K Patabendige; Diana E M Dolman; Siti R Yusof; David J Begley Journal: Neurobiol Dis Date: 2009-08-05 Impact factor: 5.996
Authors: Darlene A Kertes; Gursharan Kalsi; Carol A Prescott; Po-Hsiu Kuo; Diana G Patterson; Dermot Walsh; Kenneth S Kendler; Brien P Riley Journal: Alcohol Clin Exp Res Date: 2010-12-08 Impact factor: 3.455
Authors: Renee D Paulsen; Deena V Soni; Roy Wollman; Angela T Hahn; Muh-Ching Yee; Anna Guan; Jayne A Hesley; Steven C Miller; Evan F Cromwell; David E Solow-Cordero; Tobias Meyer; Karlene A Cimprich Journal: Mol Cell Date: 2009-07-31 Impact factor: 17.970
Authors: V K Ramanan; S L Risacher; K Nho; S Kim; S Swaminathan; L Shen; T M Foroud; H Hakonarson; M J Huentelman; P S Aisen; R C Petersen; R C Green; C R Jack; R A Koeppe; W J Jagust; M W Weiner; A J Saykin Journal: Mol Psychiatry Date: 2013-02-19 Impact factor: 15.992
Authors: Edward Y Chen; Christopher M Tan; Yan Kou; Qiaonan Duan; Zichen Wang; Gabriela Vaz Meirelles; Neil R Clark; Avi Ma'ayan Journal: BMC Bioinformatics Date: 2013-04-15 Impact factor: 3.169
Authors: Tarak Srivastava; Hongying Dai; Daniel P Heruth; Uri S Alon; Robert E Garola; Jianping Zhou; R Scott Duncan; Ashraf El-Meanawy; Ellen T McCarthy; Ram Sharma; Mark L Johnson; Virginia J Savin; Mukut Sharma Journal: Am J Physiol Renal Physiol Date: 2017-09-06
Authors: Joshua Spurrier; LaShae Nicholson; Xiaotian T Fang; Austin J Stoner; Takuya Toyonaga; Daniel Holden; Timothy R Siegert; William Laird; Mary Alice Allnutt; Marius Chiasseu; A Harrison Brody; Hideyuki Takahashi; Sarah Helena Nies; Azucena Pérez-Cañamás; Pragalath Sadasivam; Supum Lee; Songye Li; Le Zhang; Yiyun H Huang; Richard E Carson; Zhengxin Cai; Stephen M Strittmatter Journal: Sci Transl Med Date: 2022-06-01 Impact factor: 19.319