초고성능컴퓨팅 기반 건강한 고령사회 대응 빅데이터 기술 개발

Abstract

본 연구과제의 목적은 HPC(High Performance Computing) 및 데이터 기반의 질병 연구를 위한 분석 기술의 개발임. 이를 위해, 본 연구과제는 당해 연도에 다음과 같이 수행되었음. ○ 노인성 치매 대응을 위한 치매 데이터 네트워크 분석 기술 확보 ○ 치매 발병 메커니즘에 대한 가상 실험을 위한 모델링 및 시뮬레이션 기술 개발 ○ 치매 연구 지원 시스템 개발 및 확산 ○ 감염병 대응을 위한 변이 및 확산 예측 기술 개발

Ⅳ. Results of the study ○ Research and development for the biological network analysis -Development of the novel clustering and sub-network matching algorithms based on Spark platform (increase of clustering and sub-network matching execution time compared to MCL(134%) and PVF2(120%), respectively) -Identification of essential genes(15 gene) that is related to Alzheimer's disease with the gene expression data

○ Biological network modeling and simulation -Construction of the global gene regulatory network (13,141 node and 32,763 edge) -Development of boolean simulation tool (increase of simulation performance compared to BoolNet up to 10 fold)

○ Development and application of the prototype analysis platform on network and genome data for Alzheimer's disease research -Research on text mining algorithm with using CNN algorithm to extract biological entity(F-score: 78.03%) and relation(F-score: 38.8%) -Collection of somatic mutation information (54 Alzheimer's disease brain sample) -Development of the prototype analysis systems based on the big data processing platform with web interface (user’s satisfaction score: 91.2 points)

○ Development of the prediction techniques based on big data -Gathering the opinions of the experts from organization, research institute, universities through opening a seminar for cooperation with influenza and virus team in NIH, a seminar for using big data with National Health Insurance Service, 2 seminars about influenza mutation analysis techniques based on experiments and disease prediction techniques using machine learning models -Handle 2 new major consignment tasks using disease mutation and prediction core techniques -Collecting big data on disease emergence and development of processing and visualizing platform