计算机科学
数据挖掘
标记云
集合(抽象数据类型)
数据可视化
交互式可视化
可视化
情报检索
化学信息学
计算科学
可扩展性
聚类分析
分子
云计算
网格
理论计算机科学
作者
Peter Ertl,Bernhard Rohde
标识
DOI:10.1186/1758-2946-4-12
摘要
Analysis and visualization of large collections of molecules is one of the most frequent challenges cheminformatics experts in pharmaceutical industry are facing. Various sophisticated methods are available to perform this task, including clustering, dimensionality reduction or scaffold frequency analysis. In any case, however, viewing and analyzing large tables with molecular structures is necessary. We present a new visualization technique, providing basic information about the composition of molecular data sets at a single glance.A method is presented here allowing visual representation of the most common structural features of chemical databases in a form of a cloud diagram. The frequency of molecules containing particular substructure is indicated by the size of respective structural image. The method is useful to quickly perceive the most prominent structural features present in the data set. This approach was inspired by popular word cloud diagrams that are used to visualize textual information in a compact form. Therefore we call this approach "Molecule Cloud". The method also supports visualization of additional information, for example biological activity of molecules containing this scaffold or the protein target class typical for particular scaffolds, by color coding. Detailed description of the algorithm is provided, allowing easy implementation of the method by any cheminformatics toolkit. The layout algorithm is available as open source Java code.Visualization of large molecular data sets using the Molecule Cloud approach allows scientists to get information about the composition of molecular databases and their most frequent structural features easily. The method may be used in the areas where analysis of large molecular collections is needed, for example processing of high throughput screening results, virtual screening or compound purchasing. Several example visualizations of large data sets, including PubChem, ChEMBL and ZINC databases using the Molecule Cloud diagrams are provided.
科研通智能强力驱动
Strongly Powered by AbleSci AI