Question

DESeqDataSetFromMatrix

2

Entering edit mode

Gian ▴ 30

@chloe-25118

Last seen 2.3 years ago

Italy

control = factor(c(rep("Control",5),NA,NA))
affected=  factor(c(rep("Affected",7)))
library(DESeq2)
dds<-DESeqDataSetFromMatrix(
    countData=countTable,
    design =~control+affected,
    colData=data.frame(
      control=control,
      affected=affected
    ))
normCounts<-rlog(dds,blind=false)

This error coming

Error in DESeqDataSetFromMatrix(countData = countTable, design = ~control +  : 
  ncol(countData) == nrow(colData) is not TRUE

DESeqDataSetFromMatrix • 18k views

ADD COMMENT • link 3.5 years ago Gian ▴ 30

Kevin Blighe · Answer 1 · 2021-05-28

2

Entering edit mode

Kevin Blighe ★ 4.0k

@kevin

Last seen 20 days ago

Republic of Ireland

Hi,

As implied by the error message, for this command to work correctly, the following must return TRUE:

meta <- data.frame(
  control = control,
  affected = affected)
ncol(countTable) == nrow(meta)

That is, to say, the number of columns of your input raw counts table (passed to the function as countData) must be equal to the number of rows of your input sample metadata (passed to the function as colData). It also follows that the order of these (the columns and the rows) must match.

As a suggestion, I think that you may need:

dds <- DESeqDataSetFromMatrix(
  countData = countTable,
  design = ~ condition,
  colData = data.frame(
    condition = c(control, affected)))

, assuming that the elements of c(control, affected) are perfectly aligned to the columns of countTable

Kevin

ADD COMMENT • link 3.5 years ago Kevin Blighe ★ 4.0k

0

Entering edit mode

#normalizing counts
library(DESeq2)
dds<-DESeqDataSetFromMatrix(
    countData=countTable,
    design =~ condition,
    colData=data.frame(
      condition = c(control, affected)
    ))
normCounts<-rlog(dds,blind = FALSE)

When I add like this it shows this error for design Now I check that number of columns of countData is equal to the number of rows of colData

Error in DESeqDataSet(se, design = design, ignoreRank) : 
  design has a single variable, with all samples having the same value.
  use instead a design of '~ 1'. estimateSizeFactors, rlog and the VST can then be used

ADD REPLY • link updated 3.5 years ago by Kevin Blighe ★ 4.0k • written 3.5 years ago by Gian ▴ 30

2

Entering edit mode

Hi again, and thank you for replying.

control = factor(c(rep("Control",5),NA,NA))
affected = factor(c(rep("Affected",7)))

Why are there values of NA here? If you have samples whose condition is unknown, you may consider removing these samples from the dataset, or create a third 'condition' called 'Unknown'.

You may wish to try:

control = c(rep('Control', 5), 'Unknown', 'Unknown')
affected = rep('Affected', 7)
meta <- data.frame(
  condition = factor(c(control, affected),
    levels = c('Control', 'Affected', 'Unknown')))

dds <- DESeqDataSetFromMatrix(
  countData = countTable,
  design = ~ condition,
  colData = meta)