DESeq2 P/Padj values zero?

Question

[swilson]

I conducted DE analysis using DESeq2, and attempted to sort the top 100 differentially expressed genes (based on their P-values) using the following code:

dds <- DESeq(dds)
res <- results(dds, alpha = 0.05) 
resOrdered <- res[order(res$pvalue)]
topGenes <- as.data.frame(res) %>%
  tibble::rownames_to_column("GeneID") %>% 
  arrange(padj) %>% 
  select
  head(100)
topGenes

However, 22 genes have P/Padj- values of zero. Does this make any sense?

Answer 1

[jared.andrews07]

That is normal. The smallest floating point value in R is 2.225074e-308, so anything smaller than that gets rounded to 0.

[Peter]

Hello, I am now running into zeros as well with about 26 difference genes for padj and pvalues when i run deseq2. Is there a way to extend the number of floating points or should it just be read as extremely small chance due to chance.