Visual analysis of patterns in multiple amino acid mutation graphs
Visual analysis of patterns in multiple amino acid mutation graphs
复制标题
DOI：
10.1109/vast.2014.7042485
复制
发表时间：
2014-10
期刊：
影响因子：
--
通讯作者：
O. Lenz;Frank Keul;S. Bremm;K. Hamacher;T. V. Landesberger
中科院分区：
文献类型：
--
作者：
O. Lenz;Frank Keul;S. Bremm;K. Hamacher;T. V. Landesberger
蛋白质是所有生物体的基本组成部分。它们由氨基酸序列组成。与反应剂的相互作用可刺激序列中特定位置的突变。这种突变可能引发连锁反应，影响蛋白质中的其他氨基酸。需要分析链式反应，因为它们可能在药物治疗中引起不必要的副作用。突变链由有向无环图表示，其中氨基酸通过其突变依赖关系连接。由于每个氨基酸都可以单独突变，因此存在许多突变图。为了确定突变的重要影响，专家需要分析和比较这些突变图中的常见模式。然而，专家们缺乏合适的工具来实现这一目的。我们提出了一个新的系统，用于搜索和探索突变图中的频繁模式（即，基序）。我们提出了一种快速模式搜索算法，专门用于在许多突变图（即许多标记的无环有向图）中寻找生物学相关模式。我们的可视化系统允许对发现的模式进行交互式探索和比较。它可以在突变图和三维蛋白质结构中定位发现的模式。通过这种方式，可以发现潜在的有趣模式。这些模式可以作为进一步生物学分析的起点。在与生物学家的合作中，我们使用我们的方法来分析基于多个HIV蛋白酶序列的真实世界数据集。
Proteins are essential parts in all living organisms. They consist of sequences of amino acids. An interaction with reactive agent can stimulate a mutation at a specific position in the sequence. This mutation may set off a chain reaction, which effects other amino acids in the protein. Chain reactions need to be analyzed, as they may invoke unwanted side effects in drug treatment. A mutation chain is represented by a directed acyclic graph, where amino acids are connected by their mutation dependencies. As each amino acid may mutate individually, many mutation graphs exist. To determine important impacts of mutations, experts need to analyze and compare common patterns in these mutations graphs. Experts, however, lack suitable tools for this purpose. We present a new system for the search and the exploration of frequent patterns (i.e., motifs) in mutation graphs. We present a fast pattern search algorithm specifically developed for finding biologically relevant patterns in many mutation graphs (i.e., many labeled acyclic directed graphs). Our visualization system allows an interactive exploration and comparison of the found patterns. It enables locating the found patterns in the mutation graphs and in the 3D protein structures. In this way, potentially interesting patterns can be discovered. These patterns serve as starting point for a further biological analysis. In cooperation with biologists, we use our approach for analyzing a real world data set based on multiple HIV protease sequences.