Development of methods for transcript quantification anddifferential expression analysis using long-read sequencing technologies

第三代长读段测序（LRS）平台的快速发展， Pacbio和Oxford Nanopore Technologies（ONT）已经能够提高精度， 更高通量的转录物测序。长读段可以产生全长转录本 序列，克服了短读方法的大部分不确定性，以准确定义 转录本，特别是那些具有选择性剪接的基因（超过90%的人类 基因），对于其短读段测序迄今已证明是困难的。因此，LRS是 自然选择的转录变体的表达和替代的作用的研究 疾病和发育中的亚型。虽然长读技术的第一次迭代 不能产生足够的读数来量化超过最高表达的转录本，目前的 Sequel 2平台上每个SMRT细胞的测序深度高达800万次 对于更适度表达的基因来说是可靠的可定量性。此外，产量的显著增加 被报道为纳米孔。这表明LRS可能已经达到足够的吞吐量， 能够精确定量基因表达和差异表达分析。LRS 然而，转录组学数据具有在其他转录组学中不存在的特定性质 技术，例如参考转录模型的部分匹配。因此具体 需要制定量化和统计分析方法。在这个项目中，我们的目标是 为了详细描述长读数据中的数据分布， 他们的特殊阅读不确定性问题，并制定新的战略，差异表达 分析.总体目标是创建分析框架，以充分利用LRS 生物医学相关问题中同种型动力学研究的技术。

The rapid development of Third Generation, Long Read Sequencing (LRS) platforms such as Pacbio and Oxford Nanopore Technologies (ONT) have enabled increasing precision and higher-throughput sequencing of transcripts. Long reads can produce full-length transcript sequences, overcoming much of the uncertainty of short-read methods to accurately define transcripts, particularity for those genes with alternative splicing (more than 90% of human genes), for which short read sequencing has thus far proved difficult. LRS is therefore the natural choice for the study of the expression of transcript variants and of the role of alternative isoforms in disease and development. While the first iterations of the long-read technologies did not produce enough reads to quantify more than the highest expressed transcripts, the current sequencing depth of up to 8 million reads per SMRT cells on the Sequel 2 platforms promises reliable quantifiability for more modestly expressed genes. Also significant yield increases have been reported for Nanopore. This suggests that LRS may have reached sufficient throughput to enable accurate quantification of gene expression and differential expression analyses. LRS transcriptomics data have, however, specific properties that are absent in other transcriptomics technologies, such are partial matches of reference transcript models. Therefore specific methods for quantification and statistical analysis need to be developed. In this Project, we aim to characterize in detail the data distribution in long reads data, propose strategies to deal with their particular read uncertainty issues and develop new strategies for differential expression analysis. The overarching goal is to create the analytical framework to fully leverage LRS technologies for the study of isoform dynamics in relation of biomedical relevant questions.