Biological difference for selecting alignment reference to DNA vs cDNA

Question

[bharata1803]

Hello,

What is the difference of the alignment result of aligning to DNA refrence vs cDNA reference? Will these 2 results give similar inderstanding from the biological view point? What I understand from using cDNA as alignment reference is I will get the reads aligned to the transcript and if I use DNA as reference, I will get the reads aligned to the position in chromosome. I realized that using cDNA as reference, there will be no introns between exons. How do I interpret the alignment of reads to cDNA reference? This is a global question so I really appreciate your opinion. Thank you.

[jotan]

This depends on the starting material you sequenced.

If the starting material was cDNA, then aligning to genomic DNA would result in no intronic mapping. If the starting material was genomic DNA, then aligning to cDNA means you throw out data for no reason.

[bharata1803]

I think if the samples being sequenced are RNA samples (total RNA), it means there will be no introns right? Because I have done both cDNA and DNA as reference, the result concentrated on exon part.

Answer 1

[WouterDeCoster]

So you have RNA-seq data if I understood correctly. Mapping to the cDNA reference means that introns are removed in the reference and you can get continuous mapping. BUT: the cDNA reference is limited to what we know (and is properly annotated) as coding sequence, so it might (and will) be the case that parts of the reference are incomplete.

Mapping to the DNA reference means you need an aligner aware of splice junctions, such as Tophat or STAR.It's of course slightly more complicated but more complete. This is the common way of doing mapping of RNA-seq data and unless you have good reasons for doing so I would not advice to map to the cDNA genome.

However, 'more recent' mapping strategies such as kallisto use only the cDNA sequence for reference with great results, so that's also a valid approach.

The most optimal choice also depends on the organism you're studying. In the case of not-so-well annotated genomes it's better not to use cDNA based approaches.