This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Differentially expressed microRNA analysis using bowtie2-cufflink-cuffdiff

Hello all,

I am working on the differentially expressed microRNAs analysis.

I have used the bowtie2-cufflink-cuffdiff pipeline. (for the reference I downloaded from UCSC site)

I have two questions.

  1. When running cufflinks, I have ran two ways (with -g xxx.gtf option, without -g option). I know it should only makes small difference. But results look somehow different. Especially, the fpkm value as well as the status of significance. So, I will get different DEGs based on the existence of -g option. Does it make sense?
  2. Second, even though it is the fastq file of miRNAs, from the output file gene_exp.diff of cuffdiff, I can see some of reads are mapping to genes instead of miRNAs.. I may guess it is true since our experimental data can have some of mRNAs even though it intends to include microRNAs. Among 83 significantly differentially expressed ones, only 27 are mapped to microRNAs. Is it reasonable number? Or Do I miss something?
microrna deg cufflinks cuffdiff bowtie2

In addition, among the gene_exp.diff results, there are some transcripts (genes) having status of significance is "NO" because one of fpkm of samples is zero so log fold change is -inf/inf..

In my result, I have approximately 40 miRNAs having zero fpkm for Sample 2 (i.e log fold change is -inf) and approximately 90 miRNAS having zero fpkm for Sample 1 (i.e. log fold change is in) ..

Do I also have to include those miRNAs as Differentially expressed ones by recalculating the log fold change after replacing the zero to 1 or something else?

You can check once without cufflinks. Run featureCount/htseq-count providing only miRNA unique gtf, calculate counts per million, you can also normalize with DEseq2, and see if you get reasonable DEGs

0 answers

No answers yet.

Log in to answer this question.