New publication “A consensus approach to vertebrate de novo transcriptome assembly from RNA-seq data: Assembly of the duck (Anas platyrhynchos) transcriptome” accepted in Frontiers in Bioinformatics and Computational biology. In this paper Jo Morton of ADAC describes the process of generating a vertebrate transcriptome assembly. I think that most involved in this kind of work would agree that the de novo assembly of a large genome is part art, part science and part luck. For those without access to very high processing power and a lot of RAM many assemblers which combine multiple k-mers in the assembly step are slow and often fail. Jo overcame this by suggesting a method to subset the reads and generate a set of single k-mer assemblies that can then be merged or analysed separately. We noted (perhaps unsurprisingly) that the assembly statistics are very different with some methods producing very “broad” assemblies i.e. those where lots of transcripts are assembled but at the pay off of lots of short or incomplete transcripts and some are more conservative producing smaller but more “robust” assemblies. Knowing these differences any or all of the assemblies could be used for different downstream analyses. All assemblies are freely available at Figshare.