Statistical Methods for MicroRNA-Seq Experiments

项目概要/摘要 微小RNA（miRNAs）是一类小的（18-24个核苷酸）RNA，它们是基因表达的重要调节因子， 其在RNA诱导的沉默复合物（RISC）内起作用以结合mRNA并抑制翻译。改变 在miRNA表达中，已经显示出破坏整个细胞途径，实质上有助于多种 人类疾病。尽管有近25年的研究，miRNAs仍然难以测量，因为它们的短 长度、相对小的数目、序列相似性和与其他小RNA片段分离的差异。 虽然基于qPCR和微阵列的miRNA测定法仍被广泛使用，但最近的大多数研究使用小的 RNA-seq（sRNA-seq），因为它允许对isomiR（miRNA同种型）进行定量，并允许在mRNA水平上对miRNA进行测序。 鉴定新的miRNAs。从sRNA-seq数据生成的读段的处理在全局上区分了 miRNA读数和来自其他小RNA的读数，但不一定捕获miRNA的全谱 变化量处理后的sRNA-seq数据的后续统计分析仍然使用开发的方法进行 尽管sRNA-seq数据违反了这些方法的几个假设。 具体地，用于mRNA-seq数据的方法假定特征计数之间近似独立;然而， 小的miRNAs总数和少量非常高表达的miRNAs的存在导致 miRNA计数之间缺乏独立性。此外，mRNA-seq数据的标准化方法假设 或者转录的总体水平在样本之间是恒定的，或者相同数量的特征被过度， 当比较任何两个样品时，表达不足，这两个都不适用于sRNA-seq数据。的发展 解决sRNA-seq数据挑战的统计方法代表了miRNA研究的迫切需要。 我们的长期目标是通过开发针对以下方面的统计方法来推进miRNA研究： miRNA表达数据的特殊复杂性。本申请的总体目标是改善 通过开发统计方法来分析sRNA-seq数据，以应对sRNA-seq数据特有的挑战 并优于为mRNA-seq数据设计的方法。这解决了对统计方法的迫切需要 适当地分析sRNA-seq数据，这些数据现在由TCGA等大型财团定期生成， 和FANTOM。所提出的研究的基本原理是，明确解决 测量miRNAs中固有的挑战对于充分阐明miRNAs在许多人类细胞中的作用是必要的， 疾病过程。

Project Summary/Abstract MicroRNAs (miRNAs) are a class of small (18-24 nucleotide) RNAs that are essential regulators of gene expression, which act within the RNA-induced silencing complex (RISC) to bind mRNAs and suppress translation. Alterations in miRNA expression have been shown to disrupt entire cellular pathways, substantially contributing to a variety of human diseases. Despite nearly 25 years of research, miRNAs remain dicult to measure due to their short length, relatively small number, sequence similarity, and diculty to isolate from other small RNA fragments. While qPCR- and microarray-based miRNA assays are still widely used, the majority of recent studies use small RNA-seq (sRNA-seq) because it allows for the quanti cation of isomiRs (miRNA isoforms) and the possibility of identifying novel miRNAs. The processing of reads generated from sRNA-seq data globally distinguish between miRNA reads and those from other small RNAs, but do not necessarily capture the full spectrum of miRNA variation. Subsequent statistical analyses of processed sRNA-seq data are still performed using methods developed for mRNA-seq data despite the fact that sRNA-seq data violate several of the assumptions of these methods. Speci cally, methods for mRNA-seq data assume approximate independence between feature counts; however, the small total number of miRNAs and presence of a small number of very highly expressed miRNAs result in a lack of independence between miRNA counts. Additionally, normalization methods for mRNA-seq data assume either the overall level of transcription is constant across samples or an equal number of features are over- and under-expressed when comparing any two samples, neither of which hold for sRNA-seq data. The development of statistical methods that address the challenges of sRNA-seq data represents a critical need for miRNA research. Our long-term goal is to advance miRNA research by developing statistical methods that are tailored to the speci c complexities of miRNA expression data. The overall objective of this application is to improve the analysis of sRNA-seq data by developing statistical methods that account for challenges speci c to sRNA-seq data and outperform methods designed for mRNA-seq data. This addresses an urgent need for statistical methods to appropriately analyze sRNA-seq data, which are now routinely generated by large consortia such as TCGA and FANTOM. The rationale that underlies the proposed research is that methods that explicitly address the challenges inherent in measuring miRNAs are necessary to fully elucidate the role miRNAs play in many human disease processes.