Visualization Methods for 'Omics type data: Interactive exploration of High Dimensional Data

多年来，我们研究生物实体的能力一直在以指数级的速度提高，像人类基因组测序(1)这样以前既昂贵又耗时的任务已经变得更加容易实现。其中一个主要原因是下一代测序技术的发展，与旧方法相比，下一代测序技术可以以更低的成本收集更好的数据。然而，出色的数据收集并不一定与出色的知识生成相关。有了更大更复杂的数据集，充分探索和理解数据中所有有趣的结构本身就成为了一项更加耗时和资源密集的任务。这些生物数据集往往具有极高的维度，在微生物组数据集中，单个细菌物种相当于单一维度；单个样本可能包含数万或数十万种不同的细菌。大多数可视化方法在受到屏幕空间和人类理解的限制之前，一次只能处理中等数量的视觉维度。分析整个数据集对于单个用户来说是不可行的；相反，自动化方法不能利用用户的领域知识来理解哪些维度的组合是相关的(2)。虽然有很多可视化工具可用，但多样性可能会令人困惑，并且经常无法满足确切的用户需求(3)。该项目旨在开发方法，支持对这种极高维度的生物数据进行可靠和探索性的可视化分析。该研究将侧重于开发交互式和灵活的可视化方法，以支持用户获取和共享知识，并帮助生成假设。最终的可视化解决方案应满足以下一般标准：*支持极高维度数据的可视化。*能够识别和探索有趣的数据子集。*支持不同类型的组学数据分析，如微生物组学、转录组学和代谢组学。*提供支持统计数据的直观表示。此外，该项目打算在项目的所有阶段直接涉及最终用户（生物信息学家和微生物学家）。通过早期与用户交互，我们将对他们所面临的挑战有更清晰的理解，并且可以在项目进展时根据直接的用户需求和反馈来指导开发。这将通过我们与联合利华的合作来实现，该项目将通过一系列定期的短期实习，与他们的生物科学家会面。这将遵循与用户合作的循环；根据他们的输入开发方法和设计，然后根据他们的反馈迭代过程。最终的结果将是一个功能齐全的软件工具，由行业专业人士设计和测试。作为一种展示已经完成的研究的方式，也是一种可用的可视化工具。参考文献1。国家人类基因组研究所。人类基因组测序的成本。国家人类基因组研究所。【在线】2016年7月6日。https://www.genome.gov/27565109/the-cost-of-sequencing-a-human-genome/.2。joshua Krause, Aritra Dasgupta, Jean-Daniel Fekete, Enrico Bertini。SeekAView：一种导航高维数据空间的智能降维策略。2016.3。Seán I O'Donoghue, Anne-Claude Gavin, Nils Gehlenborg, David S Goodsell， Jean-Karim h斑斓，Cydney B Nielsen, Chris North, Arthur J Olson, James B Procter, David W Shattuck, Thomas Walter, Wong Bang。可视化生物数据-现在和将来。2010

Over the years, our ability to investigate biological entities has been improving at an exponential rate, tasks such as sequencing a human genome (1) that were previously costly and time consuming have become significantly more accessible. One of the main causes of this has been development of next generation sequencing technologies, which allow for superior data collection at lower costs compared to older methods.However, superior data collection doesn't necessarily correlate into superior knowledge generation. With significantly larger and more complex datasets, getting to fully explore and understanding all of the interesting structures within the data has in itself become a more time and resource intensive task.These biological datasets tend to be extremely high dimensional, within a Microbiomics dataset an individual species of bacteria equates to a single dimension; a single sample could contain tens or hundreds of thousands of different bacteria. Most visualization methods can only deal with a moderate number of visual dimensions at once, before being limited by screen space and human understanding.Analysing an entire dataset isn't feasible for a single user to do unaided; conversely automated methods cannot take advantage of the domain knowledge of the user to understand what combinations of dimensions are relevant (2). While there are a lot of visualisation tools available, the diversity can be confusing and often the exact user requirements are not met (3).This project aims to develop methods that support the reliable and explorative visual analysis of this extremely high-dimensional biological data. The research will focus on developing interactive and flexible visualisation methods that support users in gaining and sharing knowledge along with aiding hypothesis generation.The final visualization solution should address the following general criteria:* Enable visualization of extremely high dimensional data.* Enable the identification and exploration of interesting data subsets.* Support the analysis of different types of 'Omics data, such as Microbiomics, Transcriptomics and Metabolomics.* Provide visual representation of supporting statistics.Additionally, the project intends to involve the end users (bioinformaticians and microbiologists) directly throughout all stages of the project. By interacting with the users early on, we will have a much clearer understanding of the challenges that they face and can guide development based on direct user requirements and feedback as the project progresses.This will be enabled through our collaboration with Unilever, with the project going through a series of regular short placements to meet up with their bio-scientists. This will follow a cycle of collaborating with the users; developing methods and designs based on their input and then iterating the process based on their feedback. The end result of this will be a fully functional software tool, designed and tested with industry professionals. Both as a way to demonstrate the research that has been done and to be a useable visualisation tool.References1. National Human Genome Research Institute. The Cost of Sequencing a Human Genome. National Human Genome Research Institute. [Online] July 6, 2016. https://www.genome.gov/27565109/the-cost-of-sequencing-a-human-genome/.2. Josua Krause, Aritra Dasgupta, Jean-Daniel Fekete, Enrico Bertini. SeekAView: An Intelligent Dimensionality Reduction Strategy for Navigating High-Dimensional Data Spaces. 2016.3. Seán I O'Donoghue, Anne-Claude Gavin, Nils Gehlenborg, David S Goodsell, Jean-Karim Hériché, Cydney B Nielsen, Chris North, Arthur J Olson, James B Procter, David W Shattuck, Thomas Walter, Bang Wong. Vizualizing biological data - now and in the future. 2010