Logicome Profiler: Exhaustive detection of statistically significant logic relationships from comparative omics data
Logicome Profiler: Exhaustive detection of statistically significant logic relationships from comparative omics data
复制标题
DOI：
10.1371/journal.pone.0232106
复制
发表时间：
2020-05-01
期刊：
影响因子：
3.7
通讯作者：
Iwasaki, Wataru
中科院分区：
文献类型：
--
作者：
Fukunaga, Tsukasa;Iwasaki, Wataru
逻辑关系分析是一种数据挖掘方法，它从二进制矩阵数据集中全面地检测满足逻辑关系的项目三元组，例如比较基因组学中的直系同源表。由于最近的技术进步，许多二进制矩阵数据集现在正在基因组学，转录组学，表观基因组学，宏基因组学和许多其他领域用于比较目的。然而，不管假设的可解释性和逻辑关系的重要性，现有的数据挖掘方法不是基于统计假设检验的框架。这意味着，第1类和第2类错误率既不受控制，也不受估计。在这里，我们开发了Logicome Profiler，它可以从二进制矩阵数据集（Logicome意味着一些逻辑）中彻底检测统计上显着的三重逻辑关系。为了测试数据集中的所有项目三元组，同时避免假阳性，Logicome Profiler通过Bonferroni或Benjamini-Yekutieli方法调整显著性水平以进行多重检验校正。将其应用于海洋宏基因组数据集表明，Logicome Profiler可以有效地检测环境微生物和基因之间的统计上显著的三联体逻辑关系，包括尿素转运蛋白，尿素酶和光合作用相关基因之间的关系。除了组学数据分析之外，Logicome Profiler还适用于各种二进制矩阵数据集，用于寻找重要的三重逻辑关系。源代码可以在https：//github上找到。com/fukunagatsu/LogicomeProfiler。
Logic relationship analysis is a data mining method that comprehensively detects item triplets that satisfy logic relationships from a binary matrix dataset, such as an ortholog table in comparative genomics. Thanks to recent technological advancements, many binary matrix datasets are now being produced in genomics, transcriptomics, epigenomics, metagenomics, and many other fields for comparative purposes. However, regardless of presumed interpretability and importance of logic relationships, existing data mining methods are not based on the framework of statistical hypothesis testing. That means, the type-1 and type-2 error rates are neither controlled nor estimated. Here, we developed Logicome Profiler, which exhaustively detects statistically significant triplet logic relationships from a binary matrix dataset (Logicome means ome of logics). To test all item triplets in a dataset while avoiding false positives, Logicome Profiler adjusts a significance level by the Bonferroni or Benjamini-Yekutieli method for the multiple testing correction. Its application to an ocean metagenomic dataset showed that Logicome Profiler can effectively detect statistically significant triplet logic relationships among environmental microbes and genes, which include those among urea transporter, urease, and photosynthesis-related genes. Beyond omics data analysis, Logicome Profiler is applicable to various binary matrix datasets in general for finding significant triplet logic relationships. The source code is available at https://github. com/fukunagatsu/LogicomeProfiler.