Novel Data Structures And Scalable Algorithms For High Throughput Bioinformatics

测序技术的最新进展，特别是Illumina、10X基因组学、太平洋生物科学和牛津纳米孔技术的进展，正在开辟新的研究可能性和新领域。这些仪器显示了一种持续的趋势，即扩大测序吞吐量、增加读取长度和提高数据质量。与此同时，使用这些平台的成本达到了一个拐点，因此它们在生命科学领域的广泛应用变得越来越可行。然而，这种翻译需要启用生物信息学方法。*我们提出了一个生物信息学项目，以开发专门用于大型测序数据集的新数据结构，以及一个创新的RNA-SEQ组装工具，以利用最新测序平台的特性。因此，我们制定了一个具有两个目标的研究计划。*目标1.高级数据结构*创新数据类型在生物信息学应用中的价值已经被多次证明。这方面最突出的例子是使用调频索引进行快速读对齐。在这里，我们将以我们在Bloom过滤器和间隔种子方面的广泛专业知识为基础，解决生物信息学应用程序中的内存和运行时瓶颈。特别是，我们将为序列分类问题开发容错方法，其中一组高通量测序读数被分配给一组参考基因组和/或基因组座位。这一目标的结果也将支持我们提议的第二个目标的研究活动。目标2.RNA-seq组装*RNA-seq实验通常与基因组测序相结合，已被证明在研究模式物种和非模式物种的生物学方面是有用的。基于重新组装的转录组分析已经在许多项目中被证明是有用的，但它在翻译研究中的常规应用可能计算昂贵，因此需要重新考虑这个问题。使用我们将在AIM 1中开发的高级数据类型，我们将利用最新测序技术中的新信息模式，例如单细胞RNA测序(scRNA-seq)。*我们的实验室在开发、传播和维护基于先进计算方法的流行生物信息学工具方面有着既定的记录。我们将通过我们实验室的软件门户网站https://github.com/bcgsc，实施和发布我们的工具和算法，为研究社区提供广泛和及时的访问这些使能技术，并提供积极的支持。我们还将继续在生命科学领域广泛合作，应用我们的分析方法，并支持基础和应用研究项目。*本研究计划的目的是响应我们的合作者和最终用户确定的需求。最后但并非最不重要的一点是，我们希望这个项目能成为培养一批研究生和实习生/合作社学生的平台。

Latest advances in sequencing technologies, especially those from Illumina, 10X Genomics, Pacific Biosciences, and Oxford Nanopore Technologies, are opening up new possibilities and new fields of research. These instruments demonstrate a sustained trend of expanding sequencing throughput, growing read lengths, and improving data quality. In parallel, the cost of using these platforms reached an inflection point, whereby they became increasingly viable for widespread applications across life sciences. However, this translation requires enabling bioinformatics approaches.***We propose a bioinformatics project to develop novel data structures specialized for large sequencing datasets, and an innovative RNA-seq assembly tool to leverage the properties of the latest sequencing platforms. Accordingly, we have developed a research plan with two aims. ******Aim 1. Advanced Data Structures ***The value of innovative data types in bioinformatics applications has been demonstrated several times. The most prominent example of this is the use of FM-indexing for rapid read alignments. Here, we will build on our extensive expertise with Bloom filters and spaced seeds to address memory and run time bottlenecks in bioinformatics applications. Particularly, we will develop error tolerant methods for the sequence classification problem, where a set of high throughput sequencing reads are assigned to a set of reference genomes and/or genomic loci. Results of this aim will also support the research activities in the second aim of our proposal.******Aim 2. RNA-seq Assembly***RNA-seq experiments, often in combination with genome sequencing, have proven useful in studying the biology of model and non-model species. Transcriptome analysis based on de novo assembly has demonstrated utility for discovery in many projects, but its routine application in translational studies may be computationally costly, hence requires a rethinking of the problem. Using the advanced data types we will develop in Aim 1, we will leverage the new information modalities in recent sequencing technologies, such as single cell RNA sequencing (scRNA-seq). ******Our lab has an established track record of developing, disseminating, and maintaining popular bioinformatics tools built on advanced computational methods. We will implement and release our tools and algorithms through our lab's software portal at https://github.com/bcgsc, providing the research community broad and timely access to these enabling technologies, and offering active support. We will also continue to collaborate widely across life sciences domains to apply our analytical methods, and support basic and applied research projects. ******The aims of this research plan are in response to the identified needs of our collaborators and end users. Last but not the least, we expect this project to serve as a platform to train a number graduate students and interns/co-op students.*****