Building the tools for accurate RNA-seq analysis of the coding and non-coding transcriptome

越来越清楚的是，细胞功能和输出取决于多层调控关系的微调，而我们对其中大部分关系知之甚少。任何破译细胞调控网络的尝试都需要彻底了解转录组和 RNA 相互作用组。不幸的是，由于文库制备和处理流程的限制，转录组的充分表征部分仍然很窄。我们与 Lambowitz (UofTexas) 和 Abou Elela (UdeS) 团队合作，制定了 RNA-seq 文库制备方案 (TGIRT-seq)，该方案显着改进了强结构 RNA 的检测，从而可以准确检测同一样本中的所有细胞 RNA。然而，分析 RNA-seq 的计算工具仍然缺乏提供准确的定量并能够深入解释 RNA-seq 样本之间的差异。该研究计划在多个层面上解决这些挑战，以提供转录组的系统分析： -目标 1：调整参考注释以改进转录本的阅读分配。转录本定量的质量高度依赖于注释的质量。不完整的注释和多余的元素都可能导致量化不佳。为了解决这个问题，我们将在当前的注释中添加我们对不同细胞系和组织的 TGIRT-seq 数据集中的新型 RNA 的分析结果。我们还将研究使用数据驱动注释来改进读取分配。 -目标 2：改进 RNA 丰度和 RNA-RNA 相互作用的量化和可视化。目前 RNA-seq 分析流程的局限性包括对嵌入基因和多重映射读取的管理不善。我们将解决这些问题并以量化 RNA-RNA 相互作用数据集的解决方案为基础。我们还将创建数据库，提供中等大小 RNA 特征的深入可视化。上述方法以及提出的工具和方法将使 RNA 调控相互作用网络的研究成为可能。 -目标 3：构建工具以促进转录组数据及其对蛋白质组的影响的解释。我们最近描述了转录本中特定蛋白质靶向基序的动态和严格调节。这一目标建议扩大所研究的蛋白质基序的类型，首先表征线粒体靶向序列的调节，然后将它们与 ELM 数据库中定义的所有其他蛋白质基序进行比较。我们还将创建一个网络服务器，使生命科学研究人员能够自己进行此类分析。 所提出的工具和方法应该增强 RNA-seq 的分析，并在我们对基因表达和转录组功能的理解方面取得重大进展。

It is becoming increasingly clear that cellular function and output is dependent on the fine tuning of layers of regulatory relationships, most of which we poorly understand. Any attempt to decipher cell regulatory networks will require a thorough understanding the transcriptome and the RNA interactome. Unfortunately, the well characterized portion of the transcriptome still remains narrow due to limitations in both library preparation and processing pipelines. In collaboration with the Lambowitz (UofTexas) and Abou Elela (UdeS) groups, we have elaborated an RNA-seq library preparation protocol (TGIRT-seq) which significantly improves the detection of strongly structured RNAs, allowing the accurate detection of all cellular RNAs within the same sample. However, the computational tools to analyze RNA-seq are still lacking to provide an accurate quantification and to enable an insightful interpretation of differences between RNA-seq samples. This research program addresses these challenges on multiple levels to provide systematic analysis of the transcriptome: -Objective 1: To adjust reference annotations to improve read assignment to transcripts. The quality of transcript quantification is highly dependent on the quality of the annotation. Both incomplete annotations and superfluous elements can cause poor quantification. To address this issue, we will add to current annotations the results of our analysis of novel RNAs in TGIRT-seq datasets of diverse cell lines and tissues. We will also investigate the use of data-driven annotations to improve read assignment. -Objective 2: to improve the quantification and visualization of RNA abundance and RNA-RNA interactions. Current limitations of RNA-seq analysis pipelines include their poor management of embedded genes and multimapped reads. We will address these problems and build upon the solution to quantify RNA-RNA interaction datasets. We will also create databases providing insightful visualization of mid-size RNA characteristics. The methodology described above and the proposed tools and approaches will enable the study of the RNA regulatory interaction network. -Objective 3: to build tools to facilitate the interpretation of transcriptomic data and its effect on the proteome. We have recently described a dynamic and tight regulation of the inclusion of specific protein targeting motifs in transcripts. This aim proposes to widen the types of protein motifs investigated to first characterize the regulation of mitochondrial targeting sequences and then compare these to all other protein motifs defined in the ELM database. We will also create a webserver to enable researchers in the life sciences to carry out such analyses themselves. The tools and approaches proposed should enhance the analysis of RNA-seq and result in significant advances in our understanding of gene expression and transcriptome functionality.