Hey Guillermo,

I have only seen a handful of people opt for use with RnaSPAdes in publications over SPAdes or even using Trinity.

There is a presentation from one of the members of the Algorithmic Biology Lab at St Peters State University

(Which is responsible for SPAdes) Andrey Prjibelski titled "De novo transcriptome assembly Does anybody even need it?"

http://bioinformaticsinstitute.ru/sites/default/files/denovo_transcriptome_assembly_-_prjibelski_2-dec-2015.pdf

The table on page 79 which I have attached benchmarks the performance of different RNASeq assemblers and shows that RnaSPAdes is the best performer for assembling isoform variants from all the the assemblers available. The number of transcripts between SPAdes (6998transcripts) and RnaSPAdes (6916transcripts) is however really not very different.

If you are assembling a RNA virus that has sub-genomic transcription of isoforms it appears RnaSPAdes may help elucidating those isoforms but if you are looking to assemble a dataset for a metagenomics analysis and want to use the contigs downstream you will also see that Trinity performs slightly better out of all the assemblers for coverage.

My recommendation is to use both assemblers and see which one has greater amount of contigs and also use the N50 and N75 to see which dataset is more "complete". Trinity is available on most Galaxy webservers and you may have to tweak the k-mer size.

thanks so much!!