Hello,
While I am analysing microRNA data (bowtie2-htseq-count pipeline), I had a question! ( trim the adapter sequence, mapping with bowtie2) When, I aligned my sequence data, I have used the DEseq to call the DEGs.
I have two replicates for each of two comparision, which leads to the total of 4 samples. (After calling the htseq-count, as expected most of genes have 0 counts except microRNA data)
When I try to do the estimateDispersions function, it give warnings
"d <- estimateDispersions(d, method="blind", sharingMode="fit-only") "
****glm.fit: algorithm did not converge****
Anyhow, I ran the analysis with
res = nbionomTest(d, "pre", "post")
When, I looked at the result, I found that weird observation.
I cannot see any significally expressed microRNA or anything.
id baseMean baseMeanA baseMeanB foldChange log2FoldChange pval padj
21074 TMEM107 78.02900309 25.72703226 130.3309739 5.065915594 2.34082304 0.101156951 1
19610 SNORD43 667.8657405 265.6019618 1070.129519 4.029072346 2.010447711 0.146159895 1
12696 MIR4521 1.600577469 3.201154937 0 0 #NAME? 0.151698991 1
22395 UTP3 1.600577469 3.201154937 0 0 #NAME? 0.151698991 1
12382 MIR3607 1.736586708 0 3.473173417 Inf Inf 0.172783874 1
19428 SNORD101 8.134565914 2.444304911 13.82482692 5.655933863 2.49976525 0.177692193 1
19609 SNORD42B 36.92888997 16.00577469 57.85200526 3.61444581 1.853774461 0.178053555 1
This is the top of my result. (the most significant gene has p value of 0.101156951), which means there is no significant genes. (is it because, in my htseq-count matrix, most of genes has zero counts???? )
Also, there are many of microRNA genes has large counts values, however none of them are identified as DEGs.
for example, the following microRNA p values are mostly around 0.8 ~ 0.9.
pre pre post post
MIR320A 8255 6799 10681 8168
MIR30A 757 629 924 673
MIR30B 1133 440 1620 335
MIR30C1 36 27 55 19
MIR30C2 31 25 55 21
MIR30D 19044 4202 40746 5721
MIR30E 1562 519 2953 546
However,
pre pre post post
MIR3607 0 0 4 4
MIR4521 2 3 0 0
these microRNA has more significant p value comparing with the upper microRNA, which seems to strange to me.
Does it make sense? Or do I make any mistakes?
Are you actually using DESeq rather than DESeq2? If so, just stop what you're doing and switch.