I'm not following the new trends specifically for rna-seq so are there something more innovative or the old TUXEDO Tophat-cufflinks-cuffdiff etc are still good reliable and considered as standard? ( which can be easy inplemented in Galaxy)
The standard methods are reliable, although there have been efficiency, precision, and accuracy improvements that can be found out there. There are always improvements in workflows to be gained from benchmark studies and good quantitative comparisons. I have found that for the most part, the new algorithms are preferable to get the general picture, but I do not have any quantitative information to say that newer algorithms are "better" in any meaningful way beyond what is already available in the literature. Have you heard of the "pseudo-alignment" methods? They are a fascinating approach to streaming read alignment and quantitation.