第二次RNA-seq实战总结(3)-用DESeq2进行基因表达差异

DESeq2是一个用于分析基因表达差异的R包，具体操作要在R语言中运行
1.R语言安装DESeq2

>source("https://bioconductor./biocLite.R")
>biocLite("DESeq2")

2.载入基因表达量文件，添加列名

> setd("C:\Users\18019\Desk\counts")
> options(stringsAsFactors=FALSE)
> control1<-read.table("SRR957677_counts.txt",sep = "t",col.names = c("gene_id","control1"))
> head(control1)
               gene_id control11 ENSG00000000003.14_2     15762  ENSG00000000005.5_2        03 ENSG00000000419.12_2      7564 ENSG00000000457.13_3      3015 ENSG00000000460.16_5      7646 ENSG00000000938.12_2        0> control2<-read.table("SRR957678_counts.txt",sep = "t",col.names = c("gene_id","control2"))
> treat1<-read.table("SRR957679_counts.txt",sep = "t",col.names = c("gene_id","treat1"))
>treat2<-read.table("SRR957680_counts.txt",sep = "t",col.names = c("gene_id","treat2"))

3.数据整合

> ra_count <- merge(merge(control1, control2, by="gene_id"), merge(treat1, treat2, by="gene_id"))
> head(ra_count)
                 gene_id control1 control2  treat1  treat21 __alignment_not_unique  7440131  2973831 7861484 86768842            __ambiguous   976485   412543 1014239 11790513           __no_feature  1860117   768637 1289737 18120564          __not_aligned  1198545   572588 1256232 13480685        __too_lo_aQual        0        0       0       06   ENSG00000000003.14_2     1576      713    1589    1969#删除前五行>ra_count_filt <- ra_count[-1:-5,]#因为我们无法在EBI数据库上直接搜索找到ENSMUSG00000024045.5这样的基因，只能是ENSMUSG00000024045的整数，没有小数点，所以需要进一步替换为整数的形式。#将_后面的数字替换为空赋值给a>a<- gsub("\_\d", "", ra_count_filt$gene_id)#将.后面的数字替换为空赋值给ENSEMBL>ENSEMBL <- gsub("\.\d", "", a)# 将ENSEMBL重新添加到ra_count_filt1矩阵>ro.names(ra_count_filt) <- ENSEMBL
> ra_count_filt <- cbind(ENSEMBL,ra_count_filt)
> colnames(ra_count_filt)[1] <- c("ensembl_gene_id")
>head(rara_count_filt )
                ensembl_gene_id              gene_id control1 control2 treat1 treat2
ENSG00000000003 ENSG00000000003 ENSG00000000003.14_2     1576      713   1589   1969ENSG00000000005 ENSG00000000005  ENSG00000000005.5_2        0        0      0      1ENSG00000000419 ENSG00000000419 ENSG00000000419.12_2      756      384    806    984ENSG00000000457 ENSG00000000457 ENSG00000000457.13_3      301      151    217    324ENSG00000000460 ENSG00000000460 ENSG00000000460.16_5      764      312    564    784ENSG00000000938 ENSG00000000938 ENSG00000000938.12_2        0        0      0      0

4.对基因进行注释-获取gene_symbol
用bioMart对ensembl_id转换成gene_symbol

> library("biomaRt")
> library("curl")
> mart <- useDataset("hsapiens_gene_ensembl", useMart("ensembl"))
> my_ensembl_gene_id <- ro.names(ra_count_filt)
>  options(timeout = 4000000)
> hg_symbols<- getBM(attributes=c('ensembl_gene_id','hgnc_symbol',"chromosome_name", "start_position","end_position", "band"), filters= 'ensembl_gene_id', values = my_ensembl_gene_id, mart = mart)#将合并后的表达数据框ra_count_filt和注释得到的hg_symbols整合为一readcount <- merge(ra_count_filt, hg_symbols, by="ensembl_gene_id")
> head(readcount)
  ensembl_gene_id              gene_id control1 control2 treat1 treat2 hgnc_symbol chromosome_name start_position1 ENSG00000000003 ENSG00000000003.14_2     1576      713   1589   1969      TSPAN6               X      1006271092 ENSG00000000005  ENSG00000000005.5_2        0        0      0      1        TNMD               X      1005848023 ENSG00000000419 ENSG00000000419.12_2      756      384    806    984        DPM1              20       509348674 ENSG00000000457 ENSG00000000457.13_3      301      151    217    324       SCYL3               1      1698496315 ENSG00000000460 ENSG00000000460.16_5      764      312    564    784    C1orf112               1      1696620076 ENSG00000000938 ENSG00000000938.12_2        0        0      0      0         FGR               1       27612064
  end_position   band1    100639991  q22.12    100599885  q22.13     50958555 q13.134    169894267  q24.25    169854080  q24.26     27635277  p35.3#输出count表达矩阵> rite.csv(readcount, file='readcount_all.csv')
> readcount<-ra_count_filt[ ,-1:-2]
> rite.csv(readcount, file='readcount.csv')
> head(readcount)
                control1 control2 treat1 treat2
ENSG00000000003     1576      713   1589   1969ENSG00000000005        0        0      0      1ENSG00000000419      756      384    806    984ENSG00000000457      301      151    217    324ENSG00000000460      764      312    564    784ENSG00000000938        0        0      0      0

5.DEseq2筛选差异表达基因并注释(bioMart)

#载入数据（countData和colData）> mycounts<-readcount
> head(mycounts)
                control1 control2 treat1 treat2
ENSG00000000003     1576      713   1589   1969
ENSG00000000005        0        0      0      1
ENSG00000000419      756      384    806    984
ENSG00000000457      301      151    217    324
ENSG00000000460      764      312    564    784
ENSG00000000938        0        0      0      0
> condition <- factor(c(rep("control",2),rep("treat",2)), levels = c("control","treat"))
> condition
[1] control control treat   treat  
Levels: control treat
> colData <- data.frame(ro.names=colnames(mycounts), condition)
> colData
         condition
control1   control
control2   control
treat1       treat
treat2       treat

构建dds对象,开始DESeq流程

>library("DESeq2")
> dds <- DESeqDataSetFromMatrix(mycounts, colData, design= ~ condition)
> dds <- DESeq(dds)
estimating size factors
estimating dispersions
gene-ise dispersion estimates
mean-dispersion relationship
final dispersion estimates
fitting model and testing
> ddsclass: DESeqDataSet dim: 60880 4 metadata(1): version
assays(4): counts mu H cooks
ronames(60880): ENSG00000000003 ENSG00000000005 ... ENSG00000285993 ENSG00000285994
roData names(22): baseMean baseVar ... deviance maxCooks
colnames(4): control1 control2 treat1 treat2
colData names(2): condition sizeFactor#查看总体结果> res = results(dds, contrast=c("condition", "control", "treat"))
> res = res[order(res$pvalue),]
> head(res)
log2 fold change (MLE): condition control vs treat 
Wald test p-value: condition control vs treat 
Dataframe ith 6 ros and 6 columns
                        baseMean   log2FoldChange             lfcSE             stat               pvalue
                                                            
ENSG00000178691 1025.66218695436 2.83012875791025 0.225513526042636 12.5497073615672 3.98981786210676e-36ENSG00000135535 2415.77359618136 1.22406336488047 0.183431131037356 6.67314952460929  2.5037106203736e-11ENSG00000164172 531.425786834548 1.30449018960413 0.207785830749451 6.27805170785243 3.42841906697453e-10ENSG00000172239 483.998634607265 1.31701332235233 0.215453141699223 6.11275988813803 9.79226522597759e-10ENSG00000237296 53.0114998109978 2.70139282483841 0.480033904207378 5.62750422660019 1.82835684560772e-08
ENSG00000196504 3592.67315807893 1.09372324353448 0.200308218929736 5.46020153031335 4.75594407815571e-08
                                padj
                           
ENSG00000178691 3.90682965057494e-32ENSG00000135535 1.22581671973492e-07ENSG00000164172 1.11903598346049e-06ENSG00000172239 2.39714652731931e-06ENSG00000237296                   NA
ENSG00000196504 9.31404088266014e-05> summary(res)

out of 33100 ith nonzero total read count
adjusted p-value < 0.1LFC > 0 (up)       : 78, 0.24%
LFC < 0 (don)     : 15, 0.045%
outliers [1]       : 0, 0%
lo counts [2]     : 23308, 70%
(mean count < 135)
[1] see 'cooksCutoff' argument of ?results
[2] see 'independentFiltering' argument of ?results#这里可以看到有78个基因上调，15个基因下调#将分析结果输出> rite.csv(res,file="All_results.csv")

提取差异表达基因
这里我用的方法是倍差法
获取padj（p值经过多重校验校正后的值）小于0.05，表达倍数取以2为对数后大于1或者小于-1的差异表达基因

> diff_gene_deseq2 <-subset(res, padj < 0.05 & abs(log2FoldChange) > 1)
> dim(diff_gene_deseq2)
[1] 21  6
> head(diff_gene_deseq2)
log2 fold change (MLE): condition control vs treat 
Wald test p-value: condition control vs treat 
Dataframe ith 6 ros and 6 columns
                        baseMean   log2FoldChange             lfcSE             stat               pvalue
                                                            
ENSG00000178691 1025.66218695436 2.83012875791025 0.225513526042636 12.5497073615672 3.98981786210676e-36
ENSG00000135535 2415.77359618136 1.22406336488047 0.183431131037356 6.67314952460929  2.5037106203736e-11
ENSG00000164172 531.425786834548 1.30449018960413 0.207785830749451 6.27805170785243 3.42841906697453e-10
ENSG00000172239 483.998634607265 1.31701332235233 0.215453141699223 6.11275988813803 9.79226522597759e-10
ENSG00000196504 3592.67315807893 1.09372324353448 0.200308218929736 5.46020153031335 4.75594407815571e-08
ENSG00000163848 633.066990185649 1.15489622775117 0.219655131372136 5.25777030810433 1.45812478575117e-07
                                padj
                           
ENSG00000178691 3.90682965057494e-32
ENSG00000135535 1.22581671973492e-07
ENSG00000164172 1.11903598346049e-06
ENSG00000172239 2.39714652731931e-06
ENSG00000196504 9.31404088266014e-05
ENSG00000163848 0.000230253090268928#输出差异基因> rite.csv(diff_gene_deseq2,file= "DEG_treat_vs_control.csv")#用bioMart对差异表达基因进行注释> library("biomaRt")
> library("curl")
> hg_symbols<- getBM(attributes=c('ensembl_gene_id','external_gene_name',"description"),filters = 'ensembl_gene_id', values = my_ensembl_gene_id, mart = mart)
> head(hg_symbols)
  ensembl_gene_id external_gene_name
1 ENSG00000011405            PIK3C2A
2 ENSG00000100731              PCNX1
3 ENSG00000128512              DOCK4
4 ENSG00000135535              CD164
5 ENSG00000140526              ABHD2
6 ENSG00000144228              SPOPL
                                                                                                  description
1 phosphatidylinositol-4-phosphate 3-kinase catalytic subunit type 2 alpha [Source:HGNC Symbol;A:HGNC:8971]
2                                                               pecanex 1 [Source:HGNC Symbol;A:HGNC:19740]
3                                              dedicator of cytokinesis 4 [Source:HGNC Symbol;A:HGNC:19192]
4                                                           CD164 molecule [Source:HGNC Symbol;A:HGNC:1632]
5                                         abhydrolase domain containing 2 [Source:HGNC Symbol;A:HGNC:18717]
6                                       speckle type BTB/POZ protein like [Source:HGNC Symbol;A:HGNC:27934]#合并数据:res结果hg_symbols合并成一个文件> ensembl_gene_id<-ronames(diff_gene_deseq2)
> diff_gene_deseq2<-cbind(ensembl_gene_id,diff_gene_deseq2)
> colnames(diff_gene_deseq2)[1]<-c("ensembl_gene_id")
> diff_name<-merge(diff_gene_deseq2,hg_symbols,by="ensembl_gene_id")
> head(diff_name)
Dataframe ith 6 ros and 9 columns
  ensembl_gene_id         baseMean   log2FoldChange             lfcSE             stat               pvalue
                                                   
1 ENSG00000011405 1600.01408863821 1.07722909393382  0.24714564887963 4.35868120202462 1.30848557424083e-05
2 ENSG00000100731 1162.93822827396  1.0006257630015 0.214393389946423 4.66724166846545 3.05270197242525e-06
3 ENSG00000128512 368.442571635954 1.19657846347522 0.262780839813213  4.5535224878867 5.27550292947225e-06
4 ENSG00000135535 2415.77359618136 1.22406336488047 0.183431131037356 6.67314952460929  2.5037106203736e-11
5 ENSG00000140526 796.447227235737 1.05296203760187  0.23492350092969 4.48214858639031 7.38952622958053e-06
6 ENSG00000144228 293.746859588111 1.10903132755747 0.283181091639851 3.91633255290906  8.9906210067011e-05
                  padj external_gene_name
                     
1  0.00533862114290257            PIK3C2A
2    0.002491004809499              PCNX1
3  0.00319558231320097              DOCK4
4 1.22581671973492e-07              CD164
5  0.00364483675888146              ABHD2
6   0.0214722343652725              SPOPL
                                                                                                  description
                                                                                                  
1 phosphatidylinositol-4-phosphate 3-kinase catalytic subunit type 2 alpha [Source:HGNC Symbol;A:HGNC:8971]
2                                                               pecanex 1 [Source:HGNC Symbol;A:HGNC:19740]
3                                              dedicator of cytokinesis 4 [Source:HGNC Symbol;A:HGNC:19192]
4                                                           CD164 molecule [Source:HGNC Symbol;A:HGNC:1632]
5                                         abhydrolase domain containing 2 [Source:HGNC Symbol;A:HGNC:18717]
6                                       speckle type BTB/POZ protein like [Source:HGNC Symbol;A:HGNC:27934]#输出含注释的差异基因文件rite.csv(diff_name,file= "diff_gene.csv")

作者孤独巡礼_435a
链接https://.jianshu./p/4d0812195b65