De novo assembly of nanoCAGE sequences

[henning]

I am considering nanoCAGE (http://www.ncbi.nlm.nih.gov/pubmed/20543846) for non-model plant species (no or only poorly annotated genomes available) for annotation and gene expression quantification purposes. The service provider I have talked insists that nanoCAGE sequences are not meaningful without an accurate genome reference but does not give a satisfying reason why de novo assembly of the tags and quantification by alignment to the assembled short sequences would not be possible.

Does anyone have experience on using nanoCAGE without a genome reference or any insights of why this would be a bad idea [why de novo assembly would not work for sequenced 5' tags]?