Differential expression analysis of non-coding RNA-Seq data

Question

[Assa Yeroslaviz]

Hi,

I'm working on a data set from worm (C. elegans). I would like to analyse possible differential expression of ncRNA in the samples. I have five different conditions in triplicates.

for the mapping I have doewloaded the data from Ensembl - Caenorhabditis_elegans.WBcel235.ncrna.fa.

After mapping (STAR) and quantification (featureCounts), I would like to run a DE analysis using the DESeq2 or the edgeR packages.

I was wondering if the analysis can be done with similar parameters as a regular RNA-Seq analysis or do some of the parameters must be changed accordingly.

thanks in advance Assa

[h.mon]

You should map against both coding and non-coding transcripts, to avoid some reads aligning to non-coding transcripts due to lack of better alignments from the correct region.

[Assa Yeroslaviz]

This I did. In the first mapping run I have the complete gtf file containing all genes/transcripts including the ncRNA transcripts.

How can I than do a DE analysis on the ncRNA transcripts? Or should I include them in the complete analysis? When I count the features for the total RNA-Seq using featureCounts I count the genes using this command:

featureCounts -T 16 -b -a Cel.WBcel235.gtf  -t exon -g gene_id -o OUT.txt *.bam

Does it make sense to extract all the ncRNA from the gtf file and than count on transcript level?

[Assa Yeroslaviz]

something like this

featureCounts -T 15 -b -a Cel.WBcel235_subset.gtf -f -O -t exon -g transcript_id -o OUT.txt *.bam

whereas the Cel.WBcel235_subset.gtf is the file containing only the rows from the original gtf file containing ncRNA.

[Asaf]

I think you're good to go, there shouldn't be a difference from coding RNA.

[Assa Yeroslaviz]

can I use the regular gtf file for the quantification of the counts?

[Asaf]

I thought that you mapped against the ncRNA genes only. Is the gtf you are referring to alighned with the fasta file? If so then you should use it.

[Assa Yeroslaviz]

Yes, I aligned only against the ncRNA with the file mentioned above. But I don't have a gtf file for it, so I was going to use the gtf file from the RNA-Seq analysis. But when I'm using the original gtf file all I can't seems to find any counts. The assigned reads when running featureCounts is 0% for all samples.

e.g.

Process BAM file L18548_Track-45795.sorted.bam...                          
||    Single-end reads are included.                                          
||    Assign reads to features...                                             
||    Total reads : 8095454                                                   
||    Successfully assigned reads : 0 (0.0%)                                  
||    Running time : 0.07 minutes

[Asaf]

Makes sense see @h.mon remark below. The corresponding gtf file you should generate will basically contain all the contigs in your fasta file from start to end.