《外文翻译--Genome-wide Association Studies of Copy Number Variation in Glioblastoma》由会员分享，可在线阅读，更多相关《外文翻译--Genome-wide Association Studies of Copy Number Variation in Glioblastoma（4页珍藏版）》请在金锄头文库上搜索。
1、Genome-wide Association Studies of Copy Number Variation in Glioblastoma Momiao Xiong1,2 , Hua Dong1, Hoicheong Siu1, Gang Peng2, Yi Wang1 and Li Jin1 1State Key Laboratory of Genetic Engineering and MOE Key Laboratory of Contemporary Anthropology, School of Life Sciences and Institutes of Biomedica
2、l Sciences, Fudan University, Shanghai, 200433, China 2Human Genetics Center, University of Texas School of Public Health, Houston, TX 77030, USA E-mail: ABSTRACT Copy-number variation (CNV) constitutes a large proportion of total genomic variation and is increasingly recognized to be an extremely i
3、mportant risk factor for cancer. To examine the role of CNVs in glioblastoma, a genome-wide association study of CNVs in glioblastoma was conducted by assaying 221 tumor tissues and 28 normal tissues samples from primary glioblastoma multiform patients in TCGA project. CNVs were measured by the Affy
4、metrix Genome-Wide Human SNP Array 6.0 with 906,600 SNPs and more than 946,000 probes for the detection of copy number variation. CNVs were called by the modified Hidden Markov Models (HMM) and 163024 CNV loci were detected. A total of 197 CNV loci with P-value3.06*10-7 showed significant associatio
5、n with glioblastoma. We also did group association tests of CNV with glioblastoma by gene and pathway. We identified 169 genes with P-values 4.77*10-6, including oncogene, tumor suppressor genes, transcription factor and transcription activator genes, which were significantly associated with gliobla
6、stoma. We also identified 15 significantly associated pathways with glioblastoma with FDR P-values0.05. These significant pathways include Metabolism of xenobiotics by cytochrome P450, Calcium signaling pathway, Axon guidance， Colorectal cancer， Tight junction，Regulation of eIF2 pathway and Glioma.
7、Our results provide important clues for investigation of the mechanisms and drug targets of glioblastoma. Keywords- glioblastoma ; CNV; GWAS;Glioblastoma I. INTRODUCTION Glioblastoma is the most common and aggressive type of primary brain tumor in humans. Glioblastoma is located preferentially in th
8、e cerebral hemispheres. Glioblastoma arises from complex interactions between a variety of genetic, epigenetic alterations and environmental perturbations. However, the precise mechanism of glioblastoma is unknown and its survival rate is very low. Genome-wide association studies (GWAS) that focused
9、 largely on the contribution of SNPs to diseases are the current major approach to dissecting complex genetic structure of the common diseases. Although great progress in GWAS has been made, the significant SNP associations identified by GWAS only account for a few percent of the genetic variance 1.
10、 It is indispensible to find remaining genetic variants. In addition to single SNP genetic variation, copy number variation (CNV) is an important source of genetic variation. (CNV) constitutes a large proportion of total genomic variation and is increasingly recognized to be an extremely important r
11、isk factor for cancer. It is essential to perform GWAS of CNV for dissecting genetic structure of glioblastoma. The current GWAS of CNV have primarily focused on testing associations of a single CNV with a disease one at a time. Since common diseases are often caused by multiple genes and environmen
12、ts that are organized into a myriad of complex networks, to only test for association of a single CNV has limited utility 2 and is insufficient to dissect the complex genetic structure of common diseases for the following reasons. Common diseases often arise from the joint action of multiple loci wi
13、thin a gene or joint action of multiple genes within a pathway. What has been generally missing in the current GWAS is the context in which DNA variation occurs. It was reported that a gene location within a cellular network may have significant impact on the results of the given gene mutation 3. Th
14、e genetic variation occurring at multiple loci often perturbs signal, regulatory and metabolic pathways, resulting in complex changes in phenotype. To gain into insights of understanding function of CNV, in this report we propose gene and pathway-based GWAS of CNV in which a gene or a pathway is tak
15、en as a basic unit of association analysis. 2. METHODS 2.1. Test association of a single CNV with the disease DNA copy number variation often contributes to initiation and progression of cancer. We propose two statistics to formally test association of CNV with the cancer. A copy number can be viewe
16、d as a genotype. Assume that the maximum number of copies at loci is m. If we consider only gain or loss, there are three genotypes: (1) gain, (2) loss and (3) unchanged. In this case, m=3. Let TAmAAPPP,.,1=and TmPPP,1?=be vectors of allele frequencies in cases and controls, respectively. Let TAAAmA
17、APPPPdiag)(),(1=? and TmPPPPdiag=),(1?. Define the statistic for testing association of CNV with the cancer as)()(PPPPTATACNV=, where 978-1-4244-4713-8/10/$25.00 2010 IEEE+=GAAnn11 and is a generalized inverse of the matrix . The statistic CNVTis to compare differences in the frequencies of copy num
18、ber between tumor and normal tissues. Under the null hypothesis of no association of CNV with the disease, CNVT is asymptotically distributed as a central 2)(k distribution, where k is the rank of the matrix . 2.2. Gene-based Association of CNVs with Cancer. A gene-based association analysis uses a
19、gene as the basic unit of analysis. The gene-based association jointly considers all CNVs within a gene. Instead of testing the association of single CNVs with cancer, gene-based association jointly tests for the association of all the CNVs within the gene. We can observe somatic CNVs and germline C
20、NVs. To test the association of somatic CNVs we will use tumor and matched normal tissue samples. To test the association of germline CNVs, we will use blood samples of cancer patients as cases and blood samples of normal individuals as controls. However, the test statistics can be used in both case
21、s. A specific copy number for a CNV can be viewed as a genotype. For the convenience of discussion, we consider only gain or loss of copy number. There are three genotypes: (1) gain, (2) loss and (3) unchanged. We define an indicator variable for the genotype of each locus. Assume that there are k C
22、NV loci within a gene. Let TikiiTikiiYYYXXX),.,(,),.,(11=be vectors of indicator variables in cases and controls, respectively. Denote their averages as TkXXX,.,1=and TkYYY,.1=. Let)(/1jkjijiXSXU=and =kjjijiYSYV1)(/ be a linear combination of indicator variables for CNVs in the gene which is referre
23、d to as a genotype score of the gene, where S(.) denotes sampling variance. Define the test statistic: GACNVGnVSnUSVUT/ )(/ )()(2+=, where S(U) and S(V) are the estimated variance of U and V, respectively, and An and Gnbe the sample sizes of cases and controls, respectively. Then, under the null hyp
24、othesis of no association of CNVs with the cancer, CNVGTis asymptotically distributed as a central 2) 1(distribution. 2.3. Pathway-based Association of CNVs. Consider m genes within a pathway. Denote the genotype score of the j-th gene of the i-th individual by ijUand ijV in cases and controls, resp
25、ectively. Let TimiiUUU,.,1=, ,.,1mUUU = ,TimiiVVV,.,1=and TmVVV,.,1=. Let S be pooled-sample variance-covariance matrix of the genotype scores. Define the test statistic as ).()(/1/1112VUSVUnnTTGACNVP+= Under the null hypothesis of no association of pathway harboring CNVs with the cancer, 2CNVPTis a
26、symptotically distributed as a central 2)(mdistribution. 3. RESULTS 3.1. Test association of copy number loci with Glioblastoma. DNA copy number variation often contributes to initiation and progression of cancer. A copy number can be viewed as a genotype. The statistic for testing association of CN
27、Vs with glioblastoma was applied to 217 tumor and 28 normal tissues in which CNVs were profiled with Affymetrix Genome-Wide SNP Array 6.0 (906,600 SNPs and 946,000 probes for copy number detection). With a newly developed robust Hidden Markov model in our group, we detected 163,024 CNV loci. We inde
28、ntified 197 CNV loci in total with P-value3.06*10-7 showed significant association with glioblastoma. Table 1 listed significantly associated SNPs with P-values less than .1000. 110 Table 1. Significantly associated CNVs with P-values 1.00E-10 Chr Start Position End Position P-values 20 40202935 402
29、06372 6.66E-16 20 9539112 9539112 2.34E-14 20 22493112 22494453 2.34E-14 20 14835190 14835190 2.85E-14 20 14831186 14831615 1.49E-13 20 40202087 40202445 1.50E-13 20 14835230 14835333 2.86E-13 20 44538839 44540560 4.44E-13 20 40207824 40209580 4.90E-13 20 37756863 37772698 1.05E-12 20 14832703 14832
30、703 1.60E-12 20 12716527 12718370 1.96E-12 20 14897154 14897201 1.96E-12 20 43743401 43743401 9.42E-12 20 9539204 9539204 1.07E-11 20 22488619 22492962 1.07E-11 20 40212060 40220009 1.07E-11 20 14831127 14831127 2.26E-11 20 4444102 4450364 2.60E-11 8 2119465 2120782 3.01E-11 20 22495347 22505099 3.0
31、1E-11 20 33108019 33108775 6.02E-11 20 37725829 37726311 6.02E-11 20 40319262 40337470 6.02E-11 20 40862648 40867876 6.02E-11 20 9533479 9535866 6.21E-11 20 23883085 23883085 6.21E-11 20 24258855 24259342 6.21E-11 3.2. Gene and Pathway-based Association of CNVs with Cancer. A gene-based association
32、analysis uses a gene as the basic unit of analysis. The gene-based association jointly considers all CNVs within a gene. Instead of testing the association of single CNVs with cancer, gene-based association jointly tests for the association of all the CNVs within the gene. We can observe somatic CNV
33、s and germline CNVs. To test the association of somatic CNVs we will use tumor and matched normal tissue samples. To test the association of germline CNVs, we will use blood samples of cancer patients as cases and blood samples of normal individuals as controls. A specific copy number for a CNV can
34、be viewed as a genotype. For the convenience of discussion, we consider only gain or loss of copy number. Pathway based association consider pathway as the basic unit and jointly consider all CNV loci in the genes within that pathway. We define new statistic to test the association of gene and pathw
35、ay with glioma. We identified 169 genes with P-values 4.77*10-6 including oncogene BCAS1, tumor repress genes CAMTA1, APC and CSMD1, transcription factor ELF2, and transcription activator genes ETV1, CREB5 and ZHX3. Table 2 listed significantly associated genes with P-value 81000. 1. Table 2. Signif
36、icantly Associated Genes with P-value 1.0E-07 Gene P-value Gene P-value NRP1 1.00 E-17 WFDC10B 2.21E-09 SLK 1.00 E-17 VPS16 1.00E-08 AFAP1L2 2.14E-12 POLR3A 1.91E-08 C20orf26 1.59E-11 ZSCAN18 2.08E-08 WFDC11 2.55E-11 SEC23B 2.71E-08 RP5-1022P6.2 5.34E-11 BCAS1 2.87E-08 JAKMIP3 7.57E-11 KIAA1219 3.19
37、E-08 ZNF569 8.31E-10 STAU1 4.44E-08 ZNF765 9.13E-10 KIF16B 4.56E-08 SLC9A8 1.11E-09 C10orf81 5.78E-08 UQCC 1.54E-09 SH3PXD2A 6.31E-08 PLCB4 1.70E-09 CHD6 6.70E-08 CRTAC1 1.85E-09 RASSF2 7.87E-08 We also identified 15 significantly associated pathways with glioblastoma with FDR 0.05 which were listed
38、 in Table 3. Glioma is a glioblastoma causing pathway and axon guidance pathway involved in neural development. The most significant genes were NRP1 (P 17100 . 1) and SLK (P 17100 . 1) which are in the Axon guidance pathway. Metabolism of xenobiotics by cytochrome P450 and Drug metabolism - cytochro
39、me P450 are major drug metabolism pathways. Figure 1 showed the most significant eight pathways in pathway based copy number association study. The pathways which contain the same genes show overlap region in the graph. 52 important genes were included in the pathway graph. 13 genes which have p-val
40、ue 0.001 are marked as red box, 39 genes which have p-value between 0.001 and 0.05 are marked as melon box and the left 18 genes are not significant and marked as green box. The 18 genes have strong connection with significant genes, so they were kept in the graph. White large box denotes those path
41、ways which have direct connection with these 8 significant pathways. Table 3. Pathways that harbor CNVs and are associated with glioblastoma Pathway FDR Metabolism of xenobiotics by cytochrome P450 2.7E-06 Calcium signaling pathway 1.5E-06 Axon guidance 2.9E-03 Colorectal cancer 2.3E-03 Tight juncti
42、on 3.4E-03 Regulation of eIF2 pathway 6.3E-03 Double Stranded RNA Induced Gene Expression 7.0E-03 Glioma 6.6E-03 Glycan structures - biosynthesis 1 6.1E-03 Jak-STAT signaling pathway 5.6E-03 Drug metabolism - cytochrome P450 6.8E-03 Keratinocyte Differentiation pathway 7.9E-03 Telomerase RNA compone
43、nt gene hTerc Transcriptional Regulation 9.3E-03 Skeletal muscle hypertrophy is regulated via AKT/mTOR pathway 2.3E-02 BCR Signaling pathway 4.4E-02 4. DISCUSSION AND CONCLUSION DNA copy number variation is an important genetic variant. They often contribute to initiation and progression of cancer.
44、Genetic and epigenetic alternations that are likely to cause tumor formation are often organized into complex biological networks. Genetic and epigenetic risk factors individually cannot fully explain initiation and progression of cancer 6. Instead of testing only the association of single CNVs with
45、 cancer, new developed statistics were used to jointly test for the association of all the CNVs within the gene or within the pathway. The significant genes and pathways found by our method were enriched in neural development and cancer initiation function, which suggest our results provide importan
46、t clues for investigation of the mechanisms and drug targets of glioblastoma. ACKNOWLEDGMENT We thank the Cancer Genome Atlas Research Network for providing data and the members of TCGAs External Scientific Committee and the Glioblastoma Disease Working Group (http:/cancergenome.nih.gov/components/)
47、. REFERENCES 1 Altshuler D, Daly MJ, Lander ES. (2008) Genetic mapping in human disease. Science.322(5903):881-8. 2 Schadt EE, Lum PY: Thematic review series: Systems biology approaches to metabolic and cardiovascular disorders. reverse engineering gene networks to identify key drivers of complex di
48、sease phenotypes. J Lipid Res 2006; 47: 2601-2613. 3 Feldman I, Rzhetsky A, Vitkup D: Network properties of genes harboring inherited disease mutations. Proc Natl Acad Sci U S A 2008; 105: 4323-4328. 4 Joyce, P. and Tavare, S. (1995). The distribution of rare alleles. J Math Biol 33, 602-618. 5 The
49、Cancer Genome Atlas Research Network (2008) Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 455: 1061-1068. 6 Liang P, Pardee AB (2003) Analyzing differential gene expression in cancer. Nat Rev Cancer 3: 869-876. Figure 1. The genes with FDR 0.05 in eight pathways.
