Reproduction GTex Box plot

Question

[peris]

Hi

I need to regenerate the expression boxplot graph for five of mine gene of interest. I can see the expression graph in GTex portal but I cant export the image. So, I have downloaded the RNAseq data from GTex portal. Is there any tool though which I can regenerate the box plot? Thanx and regards

Answer 1

[Giovanni M Dall'Olio]

To access GTEx data you need to register first on their website.

In the data download page, you are interested in the following files:

• GTEx_Data_V6_Annotations_SampleAttributesDS.txt -> table with tissue of origin and site of each sample
• GTEx_Analysis_v6_RNA-seq_RNA-SeQCv1.1.8_gene_rpkm.gct.gz -> RPKM expression for every sample/gene (note: it's a big file)

For the boxplot, you will have to adapt the following code. It uses dplyr, tidyr and ggplot2. Hope you are familiar with these libraries :-)

# preparing data
> library(dplyr)
> library(tidyr)
> library(ggplot2)
> samples = read.delim('GTEx_Data_V6_Annotations_SampleAttributesDS.txt', sep='\t') %>%
    select(SAMPID, primary.tissue=SMTS, tissue=SMTSD)
> gtex = read.table('GTEx_Analysis_2014-01-17_RNA-seq_RNA-SeQCv1.1.8_gene_rpkm.gct', skip=2, colClasses=c('character', 'character', rep('numeric', 2921)), stringsAsFactors=F, header=T)  # go take a coffee!
> gtex.bysample = expdata %>% 
   gather(SAMPID, expression, -Name, -Description) %>% 
   mutate(SAMPID=gsub('\\.', '-', SAMPID)) %>% 
   left_join(samples)

At this point you will have a dataframe with one line for every gene and sample:

> gtex.bysample
               Name Description                  SAMPID expression primary.tissue      tissue
1 ENSG00000223972.4     DDX11L1 GTEX-N7MS-0007-SM-2D7W1    0.00000          Blood Whole Blood
2 ENSG00000227232.4      WASH7P GTEX-N7MS-0007-SM-2D7W1    2.95098          Blood Whole Blood
3 ENSG00000243485.2  MIR1302-11 GTEX-N7MS-0007-SM-2D7W1    0.00000          Blood Whole Blood
4 ENSG00000237613.2     FAM138A GTEX-N7MS-0007-SM-2D7W1    0.00000          Blood Whole Blood
5 ENSG00000268020.2      OR4G4P GTEX-N7MS-0007-SM-2D7W1    0.00000          Blood Whole Blood
6 ENSG00000240361.1     OR4G11P GTEX-N7MS-0007-SM-2D7W1    0.00000          Blood Whole Blood

You can filter by your gene and tissues of interest:

runx3.blood_breast = gtex.bysample %>% filter(Description=='RUNX3', primary.tissue %in% c("Blood", "Breast"))

You can plot it with ggplot2:

runx3.blood_breast %>%
    ggplot(aes(x=primary.tissue, y=expression)) +
        geom_boxplot()

In alternative, in NCG we also provide a summary of Gene expression for cancer genes:

[peris]

Thank you Giovanni M Dall'Olio

[Tulip Nandu]

Thanks Giovanni.

Is there a typo in

gtex.bysample = expdata %>% 
   gather(SAMPID, expression, -Name, -Description) %>% 
   mutate(SAMPID=gsub('\\.', '-', SAMPID)) %>% 
   left_join(samples)

where "expdata" should be "gtex" and this step takes forever?

[Rm]

@Giovanni M Dall'Olio

How to generate genotype specific plots for an RSID

Say rs1998081 or rs 2567619 as done in this

Answer 2

[h.mon]

R can generate box-plots, and I am certain it could create a suitable box-plot for you. But for a more specific answer, you should provide more details, e.g. a small example of how the data is organized.

Answer 3

[lkmklsmn]

In order to generate the boxplot you need both the expression and genotype data. I think the genotype data for gtex is available in dbGaP.