Very High Dimensional Visual Mining of the NCI Dataset

对于提供高维数据可视化以及分析性数据挖掘技术的新工具，学术界和商界都有很大的需求。我们认为，可视化是分析的界面，并在发现过程中提供指导。作为一个主要目标，我们将调查和评估新的可视化工具，其中一些是专有的，能够显示任意数量的维度，其中一些是专有的，能够同时显示任意数量的数据。为此，我们将使用大型公共NCI DIS化合物数据集，该数据集已经针对60个癌细胞株的电池进行了测试。除了使用该数据集进行工具评估外，较小的目标将是在数据集中发现知识。我们提出了Molconn-Z化学描述符的计算和这些描述符的组合数据挖掘。以及相关联的单元线数据。这项活动的目的是发现可能在临床环境中有用的新的化合物癌症活动模式。在接下来的第二阶段研究中，我们将把选定的可视化和分析工具集成到一个强大的集成数据挖掘包中，用于商业用途。拟议的商业应用：第一阶段提案的具体目标将使我们能够使用公开可用的NCI DIS数据集来评估高维可视化和分析工具的商业潜力，并对该数据集进行数据挖掘，以寻找潜在的新发现。

There is a significant academic and commercial need for new tools that provide high dimensional data visualizations, coupled to analytical data mining techniques. We believe that visualization is the interface to analysis and provides guidance in the discovery process. As a major aim, we will investigate and evaluate new visualization tools, some of which are proprietary, capable of displaying an arbitrary number of dimensions, some of which are proprietary, capable of displaying an arbitrary number of dimensions of data simultaneously. To do this, we will use the large public NCI DIS compound dataset that has been tested against a battery of 60 cancer cell lines. In addition to tool evaluation using this dataset, a lesser aim will be knowledge discovery in the dataset. We propose calculation of the Molconn-Z chemical descriptors and the combined data mining of these descriptors. and associated cell line data. This activity is aimed at the discovery of new compound cancer activity patterns that may be useful in a clinical setting. In a follow on Phase II research study, we will integrate the selected visualization and analytic tools into a robust integrated data mining package for commercial use. PROPOSED COMMERCIAL APPLICATIONS: The Specific Aims of this Phase I proposal will allow us to evaluate the commercial potential of high dimensional visualization and analysis tools using the publicly available NCI DIS dataset, as well as data mine this dataset for potential new discoveries.