de-novo RNA assembly and assembly of aligned RNA-seq reads

Question

[thjnant]

I am reading the great review paper by Yandel et al 2012 on Genome annotation. I came across the final paragraph:

"They (RNA-seq data) can be assembled de novo — that is, independently of the genome — using tools such as ABySS, SOAPdenovo and Trinity; the resulting transcripts are then realigned to the genome in the same way as ESTs. Alternatively, the RNA-seq data can be directly aligned to the genome using tools such as TopHat, GSNAP or Scripture followed by the assembly of alignments (rather than reads) into transcripts using tools such as Cufflinks."

I was wondering what is the difference between these two approaches? What different sorts of information do we get from the two?