DESeq Plot Error

[coralgirl]

Hi,

I'm trying to plot the scatterplot of direct vs. moderated log-ratios and I keep getting an error:

Error in plot.window(...) : need finite 'ylim' values
In addition: Warning messages:
1: In min(x) : no non-missing arguments to min; returning Inf
2: In max(x) : no non-missing arguments to max; returning -Inf

This is what I have put in:

> cdsBlind <- estimateDispersions( cds, method="blind" )
> vsd <- getVarianceStabilizedData( cdsBlind )
> mod_lfc <- (rowMeans( vsd[, conditions(cds)=="treated", drop=FALSE] ) -
+ rowMeans( vsd[, conditions(cds)=="untreated", drop=FALSE] ))
> lfc <- res$log2FoldChange
> finite <- is.finite(lfc)
> table(as.character(lfc[!finite]), useNA="always")

-Inf Inf NaN <NA>
83 212 285 0
> largeNumber <- 10
> lfc <- ifelse(finite, lfc, sign(lfc) * largeNumber)
> plot( lfc, mod_lfc, pch=20, cex=.3,
+ col = ifelse( finite, "#80808040", "red" ) )

Is it because I am running analysis on samples with no replicates? (I know - not good, but it's all that I have...). Any ideas? Thanks.

[Wolfgang Huber]

Dear Coral Girl

it looks like your mod_lfc consists of all non-finite values. What are the conditions names in your object cds? Note that in the line

mod_lfc <- (
rowMeans( vsd[, conditions(cds)=="treated", drop=FALSE] ) -
rowMeans( vsd[, conditions(cds)=="untreated", drop=FALSE] ))

you select columns in the data matrix with the names treated and untreated, and if there are no such columns, rowMeans will return NaN. In this case, you need to modify the above code according to your condition names.

Hope this helps.
Wolfgang

[coralgirl]

Hi Wolfgang,

Thank you very much - that was exactly the problem. A rookie mistake on my behalf!