Deseq error : ncol(countData) == nrow(colData) is not TRUE

Question

[robert.murphy]

I am running deseq2 on a matric of species abundance persample from a metagenomic dataset and am getting the following error:

Error in DESeqDataSetFromMatrix(countData = data, colData = metadata,  : 
  ncol(countData) == nrow(colData) is not TRUE

I get what the error is telling me but I dont understand why DESeq requiers this satisfaction. Why does the number of samples and number of genes (in this case species) need the be the same? Or am I missundstanding something here?

The code generating this error is :

ddsFullCountTable = DESeqDataSetFromMatrix(
  countData = data,
  colData = metadata,
  design = Caste_composite
)

Any help would be appriciated :)

[Sujit]

Hi, I am running DEseq for analysing a count data file generated by ht-seq. I am running the following command after creating a separate file metadata with all the information from the Count.csv. i can see that there are 8 columns including gene in the cts where as in metadata i have 7 rows. As i understand this must be equal, but how to make them equal?

Please help me fix this error.

cts <- read.csv("Count.csv", header = TRUE, sep = ",")
head(cts)

    Gene SRR58581 SRR58582 SRR58583 SRR58584 SRR58585 SRR58586
1 0610005C13Rik         29         45         35         48         59         44
2 0610006L08Rik          0          0          3          0          0          9
3 0610009B22Rik          0          1          4          1         21          1
4 0610009E02Rik        234        185        283        155        128        106
5 0610009L18Rik          0          0          5          1          0          3
6 0610010F05Rik          9          9          8          6          0         12
  SRR58587
1         44
2          0
3          4
4         91
5          0
6          7

metaData <- read.csv("metadata.csv", header = TRUE, sep = ",")
metaData
    id       dex Replicate
1 SRR58581   Treated         1
2 SRR58582   Treated         2
3 SRR58583   Treated         3
4 SRR58584   Untreated      1
5 SRR58585  Untreated       2
6 SRR58586  Untreated       3
7 SRR58587  Untreated       4

dds <- DESeqDataSetFromMatrix(countData=cts, 
                          colData=metaData, 
                          design=~dex)

> Error in DESeqDataSetFromMatrix(countData = cts, colData = metaData,
> design = ~dex) :    ncol(countData) == nrow(colData) is not TRUE

[swbarnes2]

Go look at a DESeq2 vignette, like this one:

https://bioconductor.org/packages/release/bioc/vignettes/DESeq/inst/doc/DESeq.pdf

See how neither their count data nor their metadata has row numbers, and you do? Fix that.

[Sujit]

Thank you, this was helpful. As per this link,I generated the phenodata (table) using scripts and it worked.

[i.sudbery]

Dear Sujit, this is a new question and should not be posted as an answer to an existing question. Can you please repost this in its own question.

[Sujit]

Sorry for writing here, i have followed your suggestions in other threads and that has been quite helpful. I posted the question here so that i could get your reply. Thankfully its solved now.

Answer 1

[i.sudbery]

Just to follow up what ATpoint said - your colData (here that is metadata) should contain a row for each sample in the analysis, which your countData (here that is data) should contain one column for each sample. Therefore the number of rows in colData should be the same as the number of columns in countData and both should be equal to the number of samples in the study.

Answer 2

[ATpoint]

Why does the number of samples and number of genes (in this case species) need the be the same? Or am I missundstanding something here?

It talks about colData, not the columns of your matrix. Please show the colData (metadata) and the colnames of your count matrix.

[robert.murphy]

Ahhh yes opps my bad. I was geeting cofused looking at some DESeq2 documentation about data orientation ! Thanks for point it out for me :)

[ATpoint]

No problem at all :) If you need anything else just let us know.