I have a set of RNAseq data from an experiment on mouse. I was planning on performing a standard assembly on the mouse genome but a colleague told me that it is better to do a de novo assembly. What should I do and what difference would it make?
Because current high-throughput technologies generate short-reads, I would say your colleague is right in general but not necessarily in particular.
The mouse genome is very well assembled and annotated, so a reference-based assembly is very likely to quickly answer most research questions. For instance, which genes respond to this or that treatment, are there fusion events, etc.
On the other and sometimes the interesting biology (viruses, etc.) can only be revealed by de novo assemblies. But short-reads de novo assemblies are imperfect and harder to interpret.
One day we will probably have close to full-length transcripts and then yes, de novo assemblies will be the way to go.