RNA-Seq数据分析：揭示未注释的剪接 junctions 和转录多样性

需积分: 19 68 浏览量更新于2024-09-10 收藏 1.73MB PDF 举报

"这篇文章主要探讨了RNA数据分析，特别是RNA测序(RNA-Seq)在生物学研究中的应用及其带来的挑战。RNA-Seq技术可以提供全面的转录组图像，用于完整注释和量化任何样本中所有基因及其剪接异构体。尽管如此，充分利用这种技术需要复杂的计算方法。文章通过分析大量公开的RNA-seq数据集，揭示了人类基因剪接的多样性以及未被现有注释记录的剪接接头的广泛存在。" 在当前的生物学研究中，高通量RNA测序(RNA-Seq)已经成为探索复杂生物学问题的重要工具。与传统的分子生物学技术相比，RNA-Seq具有更高的灵敏度和分辨率，能够检测到转录组的细微变化，包括基因表达水平、转录剪接变异、非编码RNA以及罕见转录事件。RNA-Seq的工作流程通常包括样品准备、测序、数据生成、数据预处理、读取比对、定量分析和功能注释等步骤。文章指出，尽管RNA-Seq技术潜力巨大，但随之而来的是对计算能力的需求增加。为了从海量的测序数据中提取有用信息，需要开发和应用各种生物信息学工具和算法，例如比对工具（如STAR、HISAT2）、转录本组装工具（如TransDecoder、Cufflinks）以及差异表达分析工具（如DESeq2、edgeR）。这些工具的使用旨在准确识别和量化基因表达，同时发现新的剪接变异和基因结构。研究人员分析了来自Sequence Read Archive (SRA)的21,504个人类RNA-seq样本，将它们比对到人类基因组上，以评估与现有基因注释的符合程度。他们发现有56,861个剪接接头（约占18.6%）在至少1000个样本中未被现有GENCODE等基因注释所包含，并且这些未注释的剪接接头表达与特定组织类型相关。这表明，RNA-seq数据可以显著扩展我们对人类基因组剪接多样性的理解，揭示大量未被充分探索的遗传变异和转录事件。此外，这项工作还强调了公共数据库如SRA在促进科研合作和数据共享方面的重要性。通过这样的大型数据分析，可以揭示在小规模研究中可能忽视的模式和趋势，进一步推动基因组学和转录组学的研究。 RNA数据分析，特别是RNA-Seq技术，对于揭示基因表达的复杂性和剪接的多样性具有重要意义。然而，它也带来了巨大的计算挑战，需要持续发展和优化计算方法来应对这些挑战。随着技术的不断进步和新工具的开发，我们可以期待更深入地理解基因功能和疾病机制，从而为精准医疗和药物研发提供新的见解。

Nellore et al. Genome Biology

(2016) 17:266

Page 3 of 14

Fig. 1 Displayed is the number of exon-exon junctions J found by Rail-RNA and other alignment protocols in at least S of the 1720 brain and universal

human reference RNA-seq samples also studied by the SEQC/MACQ-III consortium [11] (i.e., SEQC). “2 aligners” (red), “3 aligners” (green), and “4

aligners” (orange) refer to junctions we found with Rail-RNA that were also found by, respectively, 1, 2, and 3 of the alignment protocols used by SEQC

of cutoffs. For each RNA-seq junction we considered, we

also evaluated whether it appeared in annotation. We con-

sidered the following levels of evidence: (1) fully annotated

junctions; (2) separately annotated junctions (typically

exon-skipping events), where both the donor and accep-

tor sites appear in one or more junctions from annotation,

but never for the same junction; (3) alternative donor and

acceptor sites, where only either the donor or the accep-

tor site appears in one or more junctions from annotation;

and (4) novel junctions, where neither donor nor acceptor

site is found in any annotated junction.

We observed that the RNA-seq junctions most widely

expressed across samples and experiments were well doc-

umented in annotation. For example, we observed that

junctions that appeared in at least 40% of human RNA-

seq samples on the SRA (S ≥ 8000) were also present in

previous annotation at least 99.8% of the time. However,

18.6% of junctions that appeared in 1000 or more sam-

ples did not appear in annotation (Fig. 2a). Many of these

unannotated junctions are partially annotated, but 3.5% of

junctions found in more than 1000 samples do not match

any splice site from an annotated junction.

We also took an investigator-focused view of the rela-

tionship between annotation and expression. Most inves-

tigators collect only a small number of samples for their

study. We restricted attention to samples where at least

100,000 RNA-seq junctions were found to rule out obvi-

ously small RNA-seq samples and samples that were

mislabeled as RNA-seq on the SRA. In each sample, we

counted the number of instances where a read maps

across a junction. (A read mapping across two junctions

thus contributes two instances.) The total number of such

“junction overlaps” across samples is a measure of the

total expression of junctions across the SRA. Most of

the reads that map to junctions map to annotated junc-

tions (Fig. 2b). In 10,090 of a total of 10,311 samples that

meet our criterion of 100,000 junctions observed, more

than 95% of junction overlaps correspond to annotated

junctions.

This represents only the bulk coverage of junctions.

We can also consider the number of junctions observed,

regardless of coverage. In 3389 out of 10,311 samples,

we observe that fewer than 80% of junctions appear in

annotation (Fig. 2c). So while the most highly covered

junctions are well annotated, there is a large number

of junctions that are well covered across multiple sam-

ples but may not appear in any given small subset of

samples.

To explore this idea further, we investigated the poten-

tial for single studies to be the sole contributors of individ-

ual unannotated junctions. In this event, the junction may

not have been called robustly across experimental proto-

cols. Here, we considered junctions that appeared in at

least P projects instead of samples. We again broke this

calculation down by the different potential levels of evi-

dence: whether the junction was entirely novel, had an

alternative donor or acceptor, an exon skip, or whether

it was fully annotated (Fig. 3). The story at the project

level mirrors the story at the sample level: 23.4% of junc-

tions found in more than 200 of the 929 projects are not

fully annotated. So unannotated junctions recur across

independent investigations.

剩余13页未读，继续阅读

fengwanwan2017

粉丝: 0
资源: 1

RNA-Seq数据分析：揭示未注释的剪接 junctions 和转录多样性

RNA-seq数据分析实用方法(2015)

在linux中用同一个版本的R 同时安装 Seurat2 和 Seurat3的教程

RNA-combine:RNA-seq数据综合数据分析工具箱

RNA测序技术及数据分析方法

RNA-Seq数据分析：基础知识与方法

Seurat对象与基因表达分析：解读单细胞RNA数据中的生物信息

RNA序列分析基础知识与方法

基因表达数据分析：从RNA-seq到功能注释

rna-seq数据分析

rna-seq数据分析 python

最新资源