I want to sequence in 454 for microbal diversity. Is it better to sequence 16srRNA or I can sequence everything and then use a software like CARMA which uses protein encoding reads to make the taxonomy analysis. What is the preffered way? if the 16s rRNA , which software is recommended for the analysis?
If you just care about diversity and not too much about function - I would go with 16S. Pick a good primer set though. You can check how well it covers known 16S by using the Seqmatch tool at the Ribosomal Database Project.
It's also really important to define what an OTU - probably best to just use the ICOMM primer set as they've done a lot of work on determining what percent identity to use for the V6 region. With that said though their primers were designed for first gen 454 and only give something like a 50bp product.
Mothur is a good place to start for high throughput 16s analysis.
I would say that you need to choose between coverage and accuracy if you are deciding between these two approaches. Sequencing of 16S RNA gives you high degree of accuracy (if you use high quality reference database such as Silva), although coverage might be an issue (we get constantly higher diversity from generating several set of primers targeted at different parts of taxonomic tree than from using universal 16S RNA primers). When sequencing everything, you skip filtering-by-amplification step, so in theory you should have higher coverage (I'm not sure it has been tested, but that would be my intuition) although I would question accuracy, as protein databases are not that strict about taxonomic assignment (I would be skeptical about taxonomic origin of a protein sequence coming from an organism that doesn't have complete genome sequenced and assembled).
If you don't expect a large number of weird things in your sample, go for sequencing 16S RNA. For that Mothur is indeed a very nice tool and make sure you visit Mothur's wiki especially Example data analysis page.