is ok to analyse 21 bp DGE tags with tophat-cufflink pipeline?

[lzu]

Hi friends,

I have 21 bp illumina RNA-seq tags samples across multiple tissues, and the reference genome sequence.

Is it Ok to use Tophat(which implement bowtie) and cufflink pipeline (http://www.nature.com/nprot/journal/....2012.016.html) to do differential expression analysis?

[Dario1984]

It is pointless. You will not be able to map any reads across splice junctions. Use Bowtie instead.

[lzu]

It is pointless. You will not be able to map any reads across splice junctions. Use Bowtie instead.

Thanks! So after I generated SAM files of multiple samples with "Bowtie", I still could use "cuffdiff" (in cufflink) for differential expression analysis with the SAM files, gene annotation(coordinate file, gtf) and the genome sequence (genome_assembly.fasta)?

[dpryan]

Tophat actually maps to the transcriptome first, so you would still get exon-exon junctions, even with these very short reads (though you're unlikely to discover new ones, if that's important to you). So yes, you could use the normal tophat->cufflinks->cuffdiff pipeline with these sorts of reads.

[lzu]

Tophat actually maps to the transcriptome first, so you would still get exon-exon junctions, even with these very short reads (though you're unlikely to discover new ones, if that's important to you). So yes, you could use the normal tophat->cufflinks->cuffdiff pipeline with these sorts of reads.

Thanks for your suggestion. I was thinking the same way. Gene structure/alternative splicing is not important to me at the moment. I just want to measure expression variation of loci among samples(different tissues).