7.1.Translation: Ribo-seq
Last updated
Last updated
Ribo-seq(有时又称为ribosome profiling)是2009年Weissman课题组首次发表的研究细胞内蛋白翻译组的二代测序技术。这种技术选择性的对有RNA上有核糖体结合的片段进行测序，这样就能获得很多翻译组的信息。
Ribo-seq大致步骤为：
裂解细胞，富集结合着核糖体的mRNA
用核酸酶消化掉mRNA上没有核糖体保护的片段
富集核糖体，进而纯化得到有核糖体所保护的片段(ribosome protected fragment, RPF)
对RPF建库测序
生物信息学的分析
Ribo-seq数据测得的RNA片段长短与small RNA-seq相似，大约分布在25~35nt区间。由于Ribo-seq选择性的捕获正在翻译的序列，所以其数据的测序片段大多比对到基因组的CDS(coding sequence)。此外，Ribo-seq还有一个明显区别于其他RNA-seq的特点，即Ribo-seq的信号在CDS区域往往呈现3-nt的周期性(图1)。这主要因为一个codon是3 nt长，翻译过程中核糖体的移动模式也具有3 nt的周期性。
图1
Ribo-seq数据可以用来做的分析主要有以下两种:
对翻译组进行定量。Ribo-seq数据可以像RNA-seq数据一样得到表达矩阵用作差异分析。如果同样的样本还有RNA-seq的数据，那么还可以进行翻译效率的分析
ORF calling。基于ribo-seq数据，我们可以注释一些novel的ORF。
本文将介绍两个Ribo-seq处理软件(Ribocode和Ribowave)，以及一个专门用来对翻译效率进行差异分析的R包(Xtail)。我们主要介绍Ribocode和Xtail的使用，Ribowave作为补充学习。
Ribocode是一种包含Ribo-seq数据预处理、可视化、识别翻译ORF的的Ribo-seq全流程分析工具。Ribocode的分析流程如下：
Xtail是一个用于计算翻译效率（TE）和差异翻译效率的R package。计算流程如下：
RiboWave是一种基于小波变换对Ribo-seq数据的进行预处理，从而识别翻译ORF，计算翻译效率的Ribo-seq全流程分析工具。RiboWave的分析流程如下：
使用docker 启动ribo-seq所用的 Docker，进入工作目录
产生transcripts annotation文件，用于后续分析。
我们知道，一个read对应一个RPF，一个RPF对应一个核糖体，而一个核糖体有A(aminoacyl-site)，P(peptidyl-site)和E(exit-site)三个位点。我们通常用P位点的位置来代表翻译的信号。
推测P site相对于RPF的位置有不同的策略。RiboCode的做法是，对于长度相同的RPF统一设定一个offset作为P site的位置，使得这个位置可以产生周期性最强的翻译信号。这样我们就能根据一个read的长度直接推测出P site的位置。
请大家注意，这里的bam文件是位于转录本的坐标上的
Output:
wtuvb1.wtuvb1.Aligned.toTranscriptome.out.sorted.pdf：a PDF file plots the aggregate profiles of the distance from the 5'-end of reads to the annotated start codons (or stop codons), which is used for examining the P-site periodicity of RPF reads on CDS regions.
wtuvb1._pre_config.txt：The P-site config file, which defines the read lengths with strong 3-nt periodicity and the associated P-site locations for each length.
这一步可以获取有翻译数据支持的ORF。 Output:
ORF.gtf/ORF.txt:contains the information of all predicted ORFs in each transcript.
ORF_collapsed.gtf /RF_collapsed.txt ：file combines the ORFs having the same stop codon in different transcript isoforms: the one harboring the most upstream in-frame ATG will be kept.
Some column names of the result file:
"ORFcount" command which will call the HTSeq-count program. Only the reads of a given length will be counted. For those ORF with length longer than a specified value (set by "-e"), the RPF reads located in first few and last few codons can be excluded by adjusting the parameters "-f" and "-l".
The reads with length between 24-35 nt aligned on predicted ORF.The reads aligned to first 15 codons and last 5 codons of ORFs and had the length longer than 100 nt will be excluded.
Xtail是对翻译效率进行差异检验的R包，它内部调用了DESeq2来实现负二项分布检验。
使用docker 启动ribo-seq所用的 Docker(bioinfo_xtail.tar.gz)，进入工作目录
自行下载和安装
1.下载RNA-seq表达矩阵（RNA_count.txt）和Ribo-seq表达矩阵（Ribo_count.txt），在这里。
2.Xtail
Users should install DESeq2 before using Xtail: start R and enter:
Install from source package Download xtail_1.1.5-source.tar.gz;
使用docker 启动ribo-seq所用的 Docker，进入工作目录
annotation.gtf: the annotation gtf should contain start_codon and stop_codon information, eg: dmel-all-r6.18.gtf
genome.fasta: genome fasta, eg: dmel-all-chromosome-r6.18.fasta
annotation_dir: the directory for all the annotation output
scripts_dir: the directory of all the scripts in the package
annotation directory, including :
start_codon.bed: the bed file annotating start codon
final.ORFs: all identified ORFs, eg: FBtr0300105_0_31_546 where FBtr0300105 refers to the transcript, 0 refers to the reading frame relative to the start of transcript, 31 refers to the start site, 546 refers to the stop codon.
This step determines the P-site position for each Ribo-seq reads length by overlapping with the annotated start codons from previous step
Ribo_bam: secondary alignment removed to ensure one genomic position per aligned read and sorted
start_codon.bed: annotated start site start_codon.bed. It is generated in the create_annotation.sh step
out_dir: the directory of the output result, eg: GSE52799
study_name: the name of all the output file, default: test. eg: SRR1039770
scripts_dir: the directory of all the scripts in the package
查看输出
P-site directory, including :
name.psite1nt.txt: the Ribo-seq reads length and its corresponding P-sites position(= offset + 1). It may look this this :
name.psite.pdf: the PDF displaying the histogram of aggregated reads
我们收集了所有已被注释的起始密码子并将 这些起始密码子和Ribo-seq 比对上的序列进行重合，分别计算Ribo-seq序列的5’端偏离起始密码子的第一个碱基A的距离(offset)。根据 Ribo-seq测序片段长度的不同，我们进一步将Ribo-seq片段分成多个组分。在每个长度对应的组分里，作出Ribo-seq片段5’端偏离起始密码子A的距离(offset)的直方图。
每一行代表不同长度的 Ribo-seq 测序片段的直方图。该数据中，30nt的reads数目最多，在30nt长度的Ribo-seq片段中，我们可以明显的看到在距离为13nt的位点含有一个峰值(peak)。鉴于大部分核糖体会在翻译起始位点停滞较多的时间，因此对于30nt长的Ribo-seq片段，其P-site位点的定义应该代表直方图中绝大多数的核糖体，因此我们将P-site位点应该定义为峰值最高的第13个碱基(13nt)的位置。
基于Ribo-seq序列及其确定的P-site位点，将规律推广到所有Ribo-seq的片段中，直接根据Ribo-seq序列的长度推断其对应的P-site位点。根据这种方法，我们可以将每一条转录本上所有的Ribo-seq片段转化为对应的P-site位点的信号点并获得转录组水平的Ribo-seq信号轨迹（signal track）。由于是由P-site位点定义出的信号轨迹，通常也被叫做P-site信号轨迹（P-sites track）。转录本上每一个位点的信号丰度代表了有多少Ribo-seq片段对应的P-site位点落在该位置上。
This step creats the P-site track for transcripts of interests using determined P-sites position from previous step. look at transcripts from chromosome X :
查看输出
Ribo_bam
exons.gtf: a gtf file for only the exons from transcripts of interest, eg: X.exons.gtf
chromosome_size: the file including all the chromosomes and its genome size. Noted: genome can be obtained by using samtools faidx function with the input of fasta file. genome may look like this:
P-site_position: the file listing the P-site position for each read length. This file can be found in the output of previous step, eg: name.psite1nt.txt
out_dir: the directory of the output result, eg: GSE52799
study_name: the name of all the output file, default: test. eg: SRR1039770
scripts_dir: the directory of all the scripts in the package
查看生成文件
bedgraph/name directory, including :
final.psite : P-site track at transcriptome wide. It may look like this :
This step can achieve multiple functions :
denoising [denoise]
降噪:在降噪过程运用了小波变换来去除原始信号中非3-nt周期性的信号使得保留的信号均具有翻译的3-nt周期性，即P-site信号。
providing predicted p.value for each given ORF to identify its translation status [pvalue,-P]
鉴定ORF的翻译潜能：根据该转录本上经降噪后的P-sites是否富集在该ORF所在的阅读框内，判断ORF的翻译潜能。P<0.05即具有翻译活性。
providing reads density (P-site/PF P-site) for each given ORF [density,-D]
reads density = ORF上reads数/ORF长度，反映翻译水平的abundance。
providing translation efficiency (TE) estimation for each given ORF [TE,-T]
TE = 翻译水平的abundance/转录水平的abundance，反映翻译效率。
providing frameshift potential (CRF score) for each given ORF [CRF,-F]
鉴定潜在核糖体移码现象。
It might take hours to perform the analysis if the input is large. It is recommended to specify the number of CPU cores through the -p option.
Run Ribowave on example:
IMPORTANT : when estimating TE, user should input the sequenced depth of Ribo-seq and the FPKM value from paired RNA-seq
on annotated ORFs
bedgraph/name:
P-site track: output from the previous step, containing the P-site track of transcripts of interest, eg: final.psite
ORF_list: ORFs of interest ,eg : final.ORFs. It is generated in the step of create_annotation.sh
Ribo-seq sequenced depth: the sequenced depth of Ribo-seq to calculate FPKM , eg: 9012445
RNA FPKM: FPKM table. It may look like this :
name.PF_psite: the denoised signal track(PF P-sites signal track) at transcriptome wide. It looks similar as the input final psite.
including chi-square P-value information. It may look like this :
result directory, including :
name.95%.mx: RiboWave makes the prediction on the translation initiation sites and gives the final translated product output (p.value < 0.05) . It may look like this :
name.density: reads density ( PF P-site ) of given ORFs. It may look like this :
name.TE: TE of given ORFs. It may look like this :
name.CRF.final: ORFs that might experience reading frame translocation. It may look like this :
| Preprocessing | ORF Iden. | Quantification TE | Translation elongation | Integrative & visualization |
|---|---|---|---|---|
Rfoot | DeepRibo | scikit-ribo | diricore | Plastid |
RiboWaltz | FLOSS | RiboDiff | Iχnos | Rfeet |
RiboWave | ORF-RATER | riborex | PausePred | RiboGalaxy |
PRICE | Xtail | ROSE | RiboProfiling | |
RiboCode | RiboMiner | riboSeqR | ||
RiboHMM | RiboTools | |||
RibORF | RIVIT | |||
RiboTaper | Shoelaces | |||
RP-BP | tRanslatome | |||
SPECtre | ||||
TITER |
1）解释翻译效率(translation efficiency,TE)的含义。
2）利用RiboWave过程中示例文件算出TE，并画出TE的分布。
3）我们给出作业文件wtuvb2（在/home/test/rna_regulation/ribo-code/wtuvb2,或者在这里下载）,利用Ribocode，画出该样本的3nt周期性示意图，计算翻译ORF，并计算每类ORF的个数，尝试解释ORF不同类别的含义。
4）我们给出作业文件RNA_count.txt和Ribo_count_2.txt（在这里下载）利用Xtail过程计算差异翻译，并绘制火山图。
