Developing methods for curating multi-omics data

开发管理多组学数据的方法

• 批准号: 9268841
• 负责人: JUN ZHU
• 金额: $ 28.23万
• 依托单位: ICAHN SCHOOL OF MEDICINE AT MOUNT SINAI
• 依托单位国家: 美国
• 财政年份: 2015
• 资助国家: 美国
• 起止时间: 2015-06-01 至 2018-05-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/9268841
• 关键词: Address; Age; Area; Big Data to Knowledge; Biological; Biomedical Research; Cancer Patient; Clinical; Clinical Data; Computing Methodologies; Copy Number Polymorphism; Data; Data Analyses; Data Files; Data Quality; Data Set; Databases; Deposition; Disease; Epigenetic Process; Gene Expression; Generations; Genes; Genetic; Genome; Genomics; Genotype; Health; Human; Individual; Label; Lead; Lung; Maps; Measures; Metadata; Methods; Methylation; Molecular; Procedures; Proteomics; Quality Control; Regulation; Research; Resources; Sampling; Sampling Errors; Statistical Data Interpretation; The Cancer Genome Atlas; base; biological systems; blood lipid; complex biological systems; data management; data mining; data sharing; data wrangling; disease phenotype; high throughput technology; metabolomics; multiple omics; predictive modeling; statistics; transcriptomics; wasting

 DESCRIPTION (provided by applicant): Biological systems employ multiple levels of regulation that enable them to respond to genetic, epigenetic, genomic, and environmental perturbations. Advances in high throughput technologies over the past several years have enabled the generation of comprehensive data sets measuring multiple aspects of biological regulation (such as genetics, epigenetics, transcriptomics, metabolomics, glycomics, proteomics, etc.). Many databases, such as TCGA (The Cancer Genome Atlas) database and the LGRC (Lung Genome Research Consortium) database, have been created for depositing diverse types of omics data and for sharing data for public dissemination. However, data errors, including sample swapping, mis-labeling, and improper data entry, during large-scale data generation and data management are inevitable. Our preliminary results indicate that sample labeling errors frequently occur in every database we examined. Data quality control (QC) is critical for all public databases. Data errors need to be identified and corrected before data is released for data analysis and data mining. Analyzing error infested data wastes public resources. Importantly, wrong data could lead to wrong scientific conclusions. And, sample errors could have a large impact on statistic power. To maximally utilize genetic, genomic, and other omics data in public databases, it is critical to properly match different types of data pertaining to the same sample or individual before applying integrative analyses. There is an urgent need for developing methods that can identify data labeling errors in large databases and properly connect diverse types of omics data pertaining to the same individual. In respond to the Big Data to Knowledge (BD2K) initiative, we will develop computational methods to address the topic area "Data Wrangling". Here we propose to develop a sample mapping procedure called MODMatcher (Multi- Omics Data matcher) to simultaneously QC multiple types of omics data (Aim 1), and to develop a suite of predictive models based on multi omics data to identify inconsistency between clinical data and omics data (Aim 2). Our proposed methods will be used to clean data, identify and correct data annotation and metadata attribute errors in large databases, which are all within the scope of the "Data Wangling".

 描述（由适用提供）：生物系统员工多个级别的调节，使他们能够对遗传，表观遗传学，基因组和环境扰动做出反应。在过去的几年中，高通量技术的进步使得能够生成综合数据集，这些数据集测量了生物学调节的多个方面（例如遗传学，表观遗传学，转录组学，代谢组学，糖基因组学，蛋白质组学等）。已经创建了许多数据库，例如TCGA（癌症基因组图集）数据库和LGRC（肺基因组研究联盟）数据库，用于存放潜水员类型的OMICS数据和共享数据以进行公共传播。但是，在大规模数据生成和数据管理期间，数据错误，包括样本交换，错误标记和数据输入不当。我们的初步结果表明，在我们检查的每个数据库中，样本标记错误经常发生。数据质量控制（QC）对于所有公共数据库至关重要。在发布数据以进行数据分析和数据挖掘之前，需要识别和纠正数据错误。分析错误感染的数据浪费了公共资源。重要的是，错误的数据可能导致错误的科学结论。而且，样本错误可能会对统计能力产生很大的影响。为了最大化公共数据库中的遗传，基因组和其他OMICS数据，在应用集成分析之前正确匹配与同一样本或个人有关的不同类型的数据至关重要。迫切需要开发可以识别大型数据库中数据标记错误的方法，并正确连接潜水员类型的OMICS数据与同一个人有关。为了回应知识的大数据（BD2K）倡议，我们将开发计算方法来解决主题领域“数据争吵”。在这里，我们建议开发一个称为ModMatcher（多摩变数据匹配器）的样本映射程序，以简单地简单地使用QC多种类型的OMIC数据（AIM 1），并基于多OMICS数据开发一套预测模型，以识别临床数据和OMICS数据之间的不一致性（AIM 2）。我们提出的方法将用于清洁数据，识别和纠正大数据库中的数据注释和元数据属性错误，这些数据库都属于“数据wangling”的范围。