Question

Any trick to use "tximport" for converting a simple read count table ?

0

Entering edit mode

Akira Imamoto • 0

@akira-imamoto-11332

Last seen 7.7 years ago

USA/Chicago/University of Chicago

Hi,

I'm hoping to use "tximport" to convert an RNA-Seq count data file for downstream analysis such as DESeq2. But I got an error as you can see below. I think everything I have is current (R3.3.1 etc). Any input will be much appreciated !!

> datafileName <- "test.txt"  ## Replace the file name with the correct one
> table=read.delim(datafileName,header=T)
> 
> countsCol=colnames(table[,4:11])
> abundanceCol=colnames(table[,12:ncol(table)])
> RefSeqID=table[,3]
> Symbol=table[,2]
> tx2gene = data.frame(RefSeqID,Symbol)
> txin = tximport(datafileName, type="none", tx2gene = tx2gene, countsCol = countsCol, 
+                 abundanceCol=abundanceCol, lengthCol=NULL, txIdCol=NULL, 
+                 importer = read_tsv)
reading in files
1 Error in .subset2(x, i, exact = exact) : 
  attempt to select less than one element in get1index

tximport • 2.0k views

ADD COMMENT • link updated 7.7 years ago by Michael Love 41k • written 7.7 years ago by Akira Imamoto • 0

score 1 · Accepted Answer · 2016-08-26

1

Entering edit mode

Michael Love 41k

@mikelove

Last seen 6 hours ago

United States

No this is not a use case for tximport. How did you make the count matrix? You should be able to use DESeqDataSetFromMatrix() straight away.

ADD COMMENT • link 7.7 years ago Michael Love 41k

0

Entering edit mode

Thank you very much for quick reply. I was trying to find a solution to my problem with DESeqDataSetFromMatrix(). I tried but it did not take a matrix as you can see below. Only way so far was to put count columns without the row name column (in this case Symbol). What am I doing wrong?

> datafileName <- "test.txt"  ## Replace the file name with the correct one
> table=read.delim(datafileName,header=T)
> 
> table=as.matrix(table[,c(2,5:12)]) ## column2 is "GeneSymbol";columns 5-12 are raw counts
> head(table)
     Symbol     MEF258   MEF258OHT MEF284   MEF284OHT MEF285   MEF285OHT MEF286   MEF286OHT
[1,] "BC021785" "     0" "     4"  "     0" "     0"  "     0" "     0"  "     0" "      1"
[2,] "Eea1"     "  3756" "  3527"  "  4339" "  3835"  "  3951" "  3831"  "  3619" "   5891"
[3,] "Adgrg6"   "  5887" "  5151"  "  4659" "  3134"  "  4463" "  5453"  "  3299" "   6849"
[4,] "Osbpl8"   "  5712" "  7739"  "  9810" " 10208"  "  3767" "  4675"  "  2923" "   6016"
[5,] "Ttc41"    "    15" "    42"  "    24" "    25"  "    10" "    36"  "    19" "     65"
[6,] "Mars"     "  6214" "  4115"  "  4704" "  2539"  "  5166" "  5006"  "  4325" "   8964"
> 
> cellid <- rep( c("1", "2", "3", "4"), each=2) 
> cellid=as.factor(cellid)
> treatment=rep(c(0, 1), 4)
> condition=as.factor(treatment)
> condition=relevel(condition, ref="0")
> meta = data.frame(cellid, condition)
> meta
  cellid condition
1      1         0
2      1         1
3      2         0
4      2         1
5      3         0
6      3         1
7      4         0
8      4         1
> 
> d <- DESeqDataSetFromMatrix(countData=table, colData = meta, design = ~condition)
Error in `rownames<-`(`*tmp*`, value = c("Symbol", "MEF258", "MEF258OHT",  : 
  invalid rownames length

ADD REPLY • link 7.7 years ago Akira Imamoto • 0

0

Entering edit mode

Sorry! I found a solution myself for the input for DESeqDataSetFromMatrix(). Simple matrix didn't work, but adding rownames to a count-data-only table solved it.

tablex=as.matrix(table[,c(5:12)]) ## column3 is "RefSeqID";columns 5-12 are raw counts
rownames(tablex)=table[,3]

Thanks for taking time for your reply!

ADD REPLY • link 7.7 years ago Akira Imamoto • 0

1

Entering edit mode

You got it. countData should be only a matrix of counts.

The correct error message was not coming through due to some recent code additions.

I've fixed the error message in the next version to say:

Error: ncol(countData) == nrow(colData) is not TRUE

ADD REPLY • link 7.7 years ago Michael Love 41k