Error in Differential Expression Analysis with Data Set from Matrix
1
0
Entering edit mode
didemdkn • 0
@didemdkn-23211
Last seen 8 months ago

I have an expression matrix containing RSEM values for genes. I want to run DESeq2 on these file using DESeq2 package. Since the RSEM values are values like "54.87632", I multiplied all the values with 10000 to make all of them integers. Then, I uploaded both the expression data and the metadata. I checked the colnames/rownames. Then, I run the code and having the error:


dds <- DESeqDataSetFromMatrix(countData = expression,
design = ~ Group)

#then the error:

converting counts to integer mode
NAs introduced by coercion to integer rangesome variables in design formula are characters, converting to factorsError in validObject(.Object) :
invalid class “DESeqDataSet” object: NA values are not allowed in the count matrix


Since I dont have any missing value:

anyMissing(expression)
[1] FALSE


I dont know how to correct my data/what is wrong about my data. Is there anyone who have the same issue/solution for this problem?

DESeq2 DifferentialExpression Bioconductor • 288 views
1
Entering edit mode
@mikelove
Last seen 2 days ago
United States

We have a pipeline for using RSEM sample.genes.results files with DESeq2:

First use these instructions:

https://bioconductor.org/packages/release/bioc/vignettes/tximport/inst/doc/tximport.html#RSEM

Then you can use DESeqDataSetFromTximport`.

0
Entering edit mode

Thank you for yor help.

However, I don't have the RSEM output files. I just downloaded level 3 data of TCGA-BRCA study and I only have the table of expressions and patient codes. The column names are the patient names and the row names are the gene names. That's all I have. Thus, I cannot use RSEM importing. That's why I tried to use "DESeqDataSetFromMatrix".

I don't have the raw data and cannot run RSEM by using them.

Again thank you for your help. If you have any suggestion about this error, it would be very good for me. Since I don't have any former bioinformatics education, I hope I did not think wrongly about this issue.

0
Entering edit mode

You can round the estimated counts from RSEM and provide to DESeqDataSetFromMatrix. I don't have code for importing the summary data from TCGA, so you will have to do that part on your own.

It will not take into account the gene length changes that we worked on in tximport (see that publication for details).