Error while loading the data structrure in the ballgown tool

Question

[Mithil Gaikwad]

Hello, I am facing an error while loading my data structure in ballgown tool, I have tried following are the commands which i am using in my data:

>setwd("/Users/kchougul/data/testdata/Ballgown/tutorial_example")
>library(ballgown)
>library(ggplot2)
>library(gplots)
>library(genefilter)
>library(GenomicRanges)
>library(plyr)
>pheno_data = read.csv(file ="design_matrix", header = TRUE, sep = ",")
>sample_full_path <- paste("SRRXXX", "SRRYYY", "SRRZZZ",pheno_data[,1], sep = ',')

SRRXXX, SRRYYY are my ballgown folders containing respective SRRXXX.gtf and SRRYYY.gtf files

>bg = ballgown(samples=as.vector(sample_full_path),pData=pheno_data)

Tue Nov 13 13:13:27 2018
Error in ballgown(samples = as.vector(sample_full_path), pData = pheno_data) :

Something is wrong: are you missing .ctab files? Do extra files/folders (other than tablemaker output folders) match your samples/dataDir/samplePattern argument(s)?

[Kristoffer Vitting-Seerup]

You should report this as a bug on GitHub site. That is the official place to repport bugs.

But why do you want to use ballgown? Which tool have you used to quantify the RNASeq data?

[Mithil Gaikwad]

Thanks for your reply.

I have used StringTie for quantification

[Mithil Gaikwad]

Thank you very much, @kristoffer.vittingseerup and @ATpoint. I appreciate your help.

Answer 1

[Kristoffer Vitting-Seerup]

Since you have used StringTie for quantification I would suggest you use DESeq2 for differential analysis - it is a much more well established tool and there is general acceptance in the DE community that count based methods (such as DESeq2) performs better than abundance based methods (such as Ballgown).

This Bioconductor workflow will give you all the details for how to run such an analysis. Please note that although the linked workflow use Salmon quantification, the tximport() function (used for getting Salmon quantified data into R) directly supports import of StringTie data as well making the workflow directly applicable to your data.

[ATpoint]

The tximport manual covers stringtie as input, and I definitely agree to use DESeq2 simply for the reason that it is well-maintained. I think ballgown is rather in low-maintanance mode.

[Kristoffer Vitting-Seerup]

You are right! Seems Ballgown have not been updated in years!

Additionally Ballgown have (at least as far as I can tell) never been benchmarked vs any of the other standard tools (voom-limma, DESeq2 or edgeR) not even in their own paper...

[ATpoint]

Just to make things clear, I am not saying that ballgown is bad or anything. Just personally, especially given that I am not a statistician, feel saver using well-maintained tools like DESeq2.

[Kristoffer Vitting-Seerup]

Actually today i found the first benchmark of ballgown (this article) where additional file 6 can be used to test it our. I just tried and it is the absolute worst with small sample numbers! It does get better with increasing sample numbers but continue to have poor recall (True Positive Rate).

[Mithil Gaikwad]

I am new to the RNA seq analysis, so I am unable to create a character vector file i.e., files as mentioned in the tximport manual. I have created my 'files.tsv' file in which samples path SRRXXX /path/to/StringTie/results/SRRXXX/t_data.ctab SRRYYY /path/to/StringTie/results/SRRXXX/t_data.ctab . . . so on

Tell me If i am wrong in making the file Then I'll be using

>tmp <- read_tsv(files[1])
>tx2gene <- tmp[, c("t_name", "gene_name")]
>txi <- tximport(files, type = "stringtie", tx2gene = tx2gene)

Since two days i have stuck to this problem, Thank you for sparing your time.

[Kristoffer Vitting-Seerup]

Please ask this as a new question - this will enable other people with the same problem to use the solutions we will provide there.

[ATpoint]

No, please go through the tximport manual carefully and see if you find the answer there. This is a very standard thing to load files into R and I am sure the manual contains an answer plus solving these little things yourself will help you improve yourself.

[Mithil Gaikwad]

Thank you very much It worked.