GXE CYP 1 A 2 Phenotype Median CYP

＜GXE組合せ特異的＞相対リスクの例 • 遺伝素因と環境要因の相互作用の 典型例 CYP 1 A 2 Phenotype 　≦Median CYP 1 A 2 Phenotype 　　　 　＞Median Likes rare/medium meat Likes well -done meat Likes rare/medium meat NAT 2 Slow 1 1. 9 0. 9 1. 2 NAT 2 Rapid 0. 9 0. 8 1. 3 NAT 2 Slow 1 0. 9 1. 3 0. 6 NAT 2 Rapid 1. 2 1. 3 0. 9 8. 8 – Gx. Eの効果, 加算的でも乗算的でもな い • 大腸がん発症の相対リスク – ハワイでの調査 Non. Smoker （Le Marchand 2001） – 環境要因: 　喫煙、焦げた赤肉を嗜好 – 遺伝素因: 　CYP 1 A 2, NAT 2のタイプ Ever. Smoker Likes well done meat 疾患発症の相対リスクが Gx. E組合せ特異的である L. Le Marchand, JH. Hankin, LR. Wilkens, et al. Combined Effects of Well-done Red Meat, Smoking, and Rapid　N-Acetyltransferase 2 and CYP 1 A 2 Phenotypes in Increasing Colorectal Cancer Risk, Cancer Epidemiol. Biomarkers Prev 2001; 10: 1259 -1266 6

　 　Integrated Biobank and Database Tohoku Medical Megabank (TMM) is an integrated biobank having both biobank and sequence facility, which develops integrated database. Integrated Biobank Cohort Biobank Sequencing Facility Study Participants Genome & Omics Analysis Health Data Participant Laboratory Clinical Questionna Data Test Data ire Data Genome & Omics Data Tohoku Medical Megabank Integrated Database “db. TMM” Data Transfer Researchers 9

　 　Data collection and integration Cohort Data collection Specimen Laboratory Test Data Specimen Data Participant Demographic Data Cohort ID Clinical Data Physiologica Questionnair l Test Data e Data (scheduled) Hospitals (Regional Medical Information Network) Blood urine test Consent Form Physiological Test Resident Register MRI Test Data monitoring / cleaning De-idenfication Data Integration and Standarization Biobank DNA QC Sequenci ng Biobank ID Genome & Omics Data Specimen Data Investigation ID Clinical ID Health Survey Data Clinical Data Tohoku Medical Megabank Integrated Database “db. TMM” 10

　 Integrated Database “db. TMM” 　 World’s first development of Tohoku Medical Megabank Integrated Database “db. TMM” integrating health and genomic data toward genomic medicine Search Constitution (Genomic data) : 　Chr 8 41519462 (rs 515071) = TT & Health status (Lab test data) : Hb. A 1 c > 6. 2 & Lifestyle (QA) : Alcohol Drinking = Yes & Disease History (QA) : Type II Diabates = Yes Subgrouping by integrated health and genomic data 3, 000 participants 150, 000 participants 11

　 　Integrated Database “db. TMM” Ultra Fast Search of Health and Genomic “Big Data” Integrated Database “db. TMM” stores health and genomic “big data” over 2. 1 million SNV sites for 1, 070 participants Statistical Characterization of Conditionally searched population Conditionally searched How can researchers understand narrowed-down population Population w/o browsing all the variable data? 3000 participants Detection of statistically significant difference Men Smokers Tohoku Medical Megabank 150, 000 participants Statistical Characterization Statistical characterization of sub-grouped population gives us a hint for research. 12

　 　The first release of “db. TMM” Release 1. 0. 0 (Mar 31, 2016) Whole Genome Sequence Data Sex, Age SNV 2. 1 million sites 1, 070 Blood Urine Test Data Questionnaire Data (not included) Demographics, Hematological test Exercise, Alcohol (Blood cell counting) drink, Smoking, Stress, Family Immunological test structure, (Allergy test) Biochemical Test Health status, Disease history, Constitution, Urine Test Work, Sleep, Association, participants of Miyagi. Earthquake Resident Cohort (Tsunami), Women (scheduled) 13

　 　User Detailed Search Any variables, combinations, conditions with AND/OR Category Facet Cohort type, Disease classification, Omics data type Data Table Data table for Conditionally Searched population Interface for Big Data Search Graph for Conditionall y Searched Population Sex, Age, BP, Smoking, Alcohol Drinking, Statistical Disease Characteristic History for Conditionally Searched Population Detection of statistically significant difference 14

　 　Integrated Database “db. TMM” catalogue 1, 070 Alcohol drink http: //www. dist. megabank. tohoku. ac. jp/ No Yes Explanations and statistical graphs for all the variables stored in Release 1. 0. 0 of our integrated database “db. TMM” No (Constitutionally) Rarely 15

　 　 将来の方向: 知識ベース ”kb. TMM” 東北メディカルメガバンク 知識ベース “kb. TMM”(in prep) RDF triple store Clinical Knowledge Literature Knowledge Associations Gx. E Interactions Knowledge extraction Participants Subgroup Large-scale Association Analysis Gx. E Interactions Analysis -log p G G E Participants Subgroup using high-dimension feature vectors T 150 K participants Health Survey Phenotype Data Genome Data, (Clinical Data) Omics Data E Integrated Database “db. TMM” 16

インテリジェント・バイオバンクの概念と ゲノム医療実現への寄与 　 東北メディカル・メガバンク 知識ベース 　“kb. TMM” Physicians Clinical Knowledge Research Product Literature Knowledge Gx. E Interactions Associations Research Community Data Sharing Participants Subgroup 人 知能、機械発見 G E T Genome Data Health Survey Data, Omics Data Phenotype Data (Clinical Data) 統合データベース “db. TMM” インテリジェント・バイオバンク Ro. R Participants 17

　 　GE-WAS 解析フロー T対象 Gi ゲノムデータ 計算 Ej 環境データ T症例 Ej T 疾患フェノタイプ Gi aa G a. A AA i 0 aa N 1 ija. A N 3 ij. AA N 5 ij Ej 6 0 1 N 1 ij. N 2 ij. N 3 ij. N 4 ij. N 5 ij. N ij 1 N 2 ij Gi N 4 ij コクラン-マンテルヘンツェル検定 N 6 ij -log(p-value) GE-wide Analysis G・E クラスタリング Ej 3 D プロット G Ej p-value E -log(p-value) Gi 21

　 　GE-WAS 解析フロー 2, 000 SNVs G 9番染色体 i 499, 738 SNVs T対象 rs 6251588* Ej 検体検査項目 41 因子 (正常/異常) T アレルギー性鼻炎: 167 件 コントロール: 901 件 計算 T症例 Ej Gi aa G a. A AA i 0 aa N 1 ija. A N 3 ij. AA N 5 ij Ej 6 0 1 N 1 ij. N 2 ij. N 3 ij. N 4 ij. N 5 ij. N ij 1 N 2 ij 41 因子 To. MMo 1 KJPN Gi N 4 ij コクラン-マンテルヘンツェル検定 N 6 ij 1, 237 SNVs -log(p-value) GE-wide Analysis 41 因子 Ej G・E クラスタリング 3 D プロット G Ej p-value E -log(p-value) *Sakashita, Clin Exp Allergy, 2008 266 SNVs (-log > 0. 30) Gi 22

　 　アレルギー性鼻炎のGx. E ランドスケーププロット -log p=26. 7 G: rs 2381416 E: カンジダ 花粉症(Schröder 2016) -log p=35. 7 -log p=43. 4 G: rs 549716210 E: イヌ皮屑 G: rs 549868264 E: ミルク rs 186758850 rs 192922406 rs 549716210 rs 2381416 rs 549868264 rs 146693626 rs 146898930 rs 568525828 rs 191520260 rs 76741691 E ハンノキ ブタクサ カンジダ イヌ皮屑 （フケ） ミルク 24