6.3.Ribo-seq

1) workflow

2) 背景介绍

Ribo-seq原理

Ribo-seq是2009年Weissman课题组首次发表的研究细胞内蛋白翻译图谱的新型二代测序技术,用来描述全基因组水平蛋白质的翻译情况。主要是选择性捕捉80S核糖体及其结合的RNA片段而定位核糖体所位于的RNA的位置。具体步骤为:在细胞裂解物中富集多聚核糖体(polysome);将多聚核糖体用核酸酶(RNA nuclease)消化成单核糖体(monosome);选择性的收集和富集80S核糖体并经纯化得到80S核糖体所保护的RNA片段。在此过程中,将80S核糖体保护的RNA片段进行下一步构建文库和测序(图1)。最后,通过生物信息学的分析获得细胞当前状态下的翻译图谱 。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是特异性描述细胞的翻译组,因此其数据的测序片段大多比对到基因组的CDS区域(coding region)。此外,Ribo-seq还有一个明显区别于其他RNA-seq的特点,即Ribo-seq的序列在CDS区域往往呈现3-nt的周期性(图1)。这主要依赖于翻译过程中核糖体通常以3-nt的周期进行移动。

图1

3) 数据处理

download scripts from https://github.com/lulab/Ribowave

(1) Requirements

software:

R, bedtools v2.25.0

R packages:

reshape, ggplot2, rhdf5, methods, wmtsa, parallel

(2) Pre-processing

0. create annotation

scripts/create_annotation.sh -G annotation_fly/dmel-all-r6.18.gtf -f annotation_fly/dmel-all-chromosome-r6.18.fasta -o annotation_fly -s scripts;

input files

  1. : the annotation gtf should contain start_codon and stop_codon information,eg: dmel-all-r6.18.gtf

  2. : genome fasta ,eg: dmel-all-chromosome-r6.18.fasta

  3. : the directory for all the annotation output

  4. : the directory of all the scripts in the package

output files

annotation directory, including : 1. start_codon.bed : the bed file annotating start codon 2. 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.

1. P-site determination

核糖体上具有一系列与蛋白质合成有关的结合位点与催化位点,分别为A位点(aminoacyl-site,A-site),P位点(peptidyl-site,P-site)和E位点(exit-site,E- site)先后与tRNA发生结合。P位点是肽段翻译延长的主要场所,在该位点上tRNA将携带的氨基酸移交给旁边的肽段从而使得肽段序列发生延长。为了能够更加明显的观察到3-nt的周期性,在处理Ribo-seq数据时我们参考之前已发表的方法,对每一条Ribo-seq比对上的测序片段转换为其对应的P-site位点。

This step determines the P-site position for each Ribo-seq reads length by overlapping with the annotated start codons from previous step

scripts/P-site_determination.sh -i GSE52799/SRR1039770.sort.bam -S annotation_fly/start_codon.bed -o GSE52799 -n SRR1039770 -s scripts;

input files

  1. : secondary alignment removed to ensure one genomic position per aligned read and sorted

  2. annotation :

    : annotated start site start_codon.bed. It is generated in the create_annotation.sh step.

  3. : the directory of the output result, eg: GSE52799

  4. : the name of all the output file, default: test. eg: SRR1039770

  5. : the directory of all the scripts in the package

output files

P-site directory, including : 1. name.psite1nt.txt : the Ribo-seq reads length and its corresponding P-sites position(= offset + 1). It may look this this :

30 13
  1. 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。

2. Generating P-site track

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 :

scripts/create_track_Ribo.sh -i GSE52799/SRR1039770.sort.bam -G annotation_fly/X.exons.gtf -g annotation_fly/genome -P GSE52799/P-site/SRR1039770.psite1nt.txt -o GSE52799 -n SRR1039770 -s scripts;

input files

    1. : a gtf file for only the exons from transcripts of interest, eg: X.exons.gtf

  1. : 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:

    2L 23513712
    2R 25286936
    3L 28110227
    3R 32079331
  2. P-site:

    : 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

  3. : the directory of the output result, eg: GSE52799

  4. : the name of all the output file, default: test. eg: SRR1039770

  5. : the directory of all the scripts in the package

output files

  1. bedgraph/name directory, including :

    final.psite : P-site track at transcriptome wide. It may look like this :

    FBtr0070533 0,0,0,0,0,0,0,0,0,0,0,0,6,2,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,3,1,0,0,0,0,1,0,2,1,0,0,0,0,0,0,4,8,0,0,3,0,5,12,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
    FBtr0073886 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,2,0,25,0,0,0,0,0,0,0,0,0,0,0,0,0
    FBtr0070604 0,0,0,0,0,0,0,0,0,0,0,0,59,6,0,1,0,0,2,6,1,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
    FBtr0070603 0,0,0,0,0,0,0,0,0,0,0,0,75,2,7,10,7,2,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,0,0,0,0,3,3,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0

(3) Main function

3. RiboWave

This step can achieve multiple functions :

  1. 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PF P-site信号轨迹。

  1. 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。

  1. 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。

  1. 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。

  1. 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:

Denoise the P-site track

mkdir -p Ribowave;
scripts/Ribowave -a GSE52799/bedgraph/SRR1039770/final.psite -b annotation_fly/final.ORFs -o GSE52799/Ribowave -n SRR1039770 -s scripts -p 8;

Identifying translated ORF

mkdir -p Ribowave;
scripts/Ribowave -P -a GSE52799/bedgraph/SRR1039770/final.psite -b annotation_fly/final.ORFs -o GSE52799/Ribowave -n SRR1039770 -s scripts -p 8;

Estimating abundance

mkdir -p Ribowave;
scripts/Ribowave -D -a GSE52799/bedgraph/SRR1039770/final.psite -b annotation_fly/final.ORFs -o GSE52799/Ribowave -n SRR1039770 -s scripts -p 8;

Estimating TE

IMPORTANT : when estimating TE, user should input the sequenced depth of Ribo-seq and the FPKM value from paired RNA-seq

mkdir -p Ribowave;
scripts/Ribowave -T 9012445 GSE52799/mRNA/SRR1039761.RPKM -a GSE52799/bedgraph/SRR1039770/final.psite -b annotation_fly/final.ORFs -o GSE52799/Ribowave -n SRR1039770 -s scripts -p 8;

Calculating frameshift potential

on annotated ORFs

mkdir -p Ribowave;
awk -F '\t' '$3=="anno"' annotation_fly/final.ORFs > annotation_fly/aORF.ORFs;
scripts/Ribowave -F -a GSE52799/bedgraph/SRR1039770/final.psite -b annotation_fly/aORF.ORFs -o GSE52799/Ribowave -n SRR1039770 -s scripts -p 8;

Multiple functions

mkdir -p Ribowave;
scripts/Ribowave -PD -T 9012445 GSE52799/mRNA/SRR1039761.RPKM -a GSE52799/bedgraph/SRR1039770/final.psite -b annotation_fly/final.ORFs -o GSE52799/Ribowave -n SRR1039770 -s scripts -p 8;

input files

  1. bedgraph/name:

    : output from the previous step, containing the P-site track of transcripts of interest, eg: final.psite

  2. : ORFs of interest ,eg : final.ORFs. It is generated in the step of create_annotation.sh

  3. : the sequenced depth of Ribo-seq to calculate FPKM , eg: 9012445

  4. : FPKM table. It may look like this :

FBtr0100871 22262
FBtr0070604 18682
FBtr0100231 14746.5
FBtr0100874 14024.5
FBtr0100864 11475.6

output files

  1. name.PF_psite : the denoised signal track(PF P-sites signal track) at transcriptome wide. It looks similar as the input final psite.

  2. including chi-square P-value information. It may look like this :

    column1-4: basic information about the ORF
    column5: reads coverage within the ORF
    column6: P-value predicted by RiboWave
    column7: Values estimating the relative abundance of PF P-sites outside of the studied ORF
    column8: Reads intensity at the current start codon

    result directory, including :

  3. 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 :

    FBtr0070007_2_93_1028
    FBtr0070008_1_128_943
    FBtr0070025_2_135_1094
  4. name.density : reads density ( PF P-site ) of given ORFs. It may look like this :

    column1-4: basic information about the ORF
    column5: number of PF P-sites in transcript
    column6: number of PF P-sites in given ORF
    column7: density of PF P-sites in given ORF
  5. name.TE : TE of given ORFs. It may look like this :

    column1: transcript
    column2: ORF
    column3: TE
  6. name.CRF.final : ORFs that might experience reading frame translocation. It may look like this :

    column1: ORF
    column2: start of frameshift
    column3: stop of frameshift
    column4: PF P-sites' reading frames after the change point ,eg: 2_2,0_1 where 2_2 refers to continuous two PF P-sites of frame 2 followed by continuous one PF P-sites of frame 0.
    column5: Relative position of PF P-sites after the shift ,eg : 1413,1440;1789 where 1413,1440 corresponds to the exact position of 2_2 within the transcript. Discontinuity in the reading frame is separated by ;
    column6: CRF score describing the potential of frameshift

4) 数据库推荐

http://lulab.life.tsinghua.edu.cn/postar/