Warning message of NAs introduced by coercion after nbinomWaldTest
1
0
Entering edit mode
Richelle • 0
@c8796f9b
Last seen 6 weeks ago
Sweden

Hello,

I have been having an error showing this:

69 rows did not converge in beta, labelled in mcols(object)\$betaConv. Use larger maxit argument with nbinomWaldTest


so I replaced the code with this:

dds <- DESeqDataSetFromMatrix(countData=cells2, colData = info2, design = m1)
dds <- estimateSizeFactors(dds)
nc <- counts(dds, normalized=TRUE)
filter <- rowSums(nc >= 10) >= 2
dds <- dds[filter,]
dds <- estimateDispersions(dds)
dds <- nbinomWaldTest(dds, maxit=5000000000, modelMatrix=m1)


I have still been getting rows that don't converge that is why my maxit reached this big. Is this big maxit okay? Anyway, so I don't get the error anymore on rows not converging but I get this warning message instead:

In fitBeta(ySEXP = ySEXP, xSEXP = xSEXP, nfSEXP = nfSEXP, alpha_hatSEXP = alpha_hatSEXP,  :
NAs introduced by coercion to integer range


I tried to search online but couldn't find a similar warning. What does this mean? Should I change anything or I can proceed with my analysis?

Thank you!

DESeq2 • 97 views
0
Entering edit mode
@mikelove
Last seen 3 hours ago
United States

I wouldn’t bother past maxit of 1000. Can you examine the rows that still don’t converge and see if they can be filtered? Last resort you can just exclude those unconverged results from output.

0
Entering edit mode

Great! Thank you, Michael :)