There's an idea that Dijkstra was always harping on about, among the various other things he harped on about: separation of concerns. Libraries to help with web development have no reason to be bioinformatics specific. It is counterproductive to make them so, since then you only have the relatively small bioinformatics community, where the average level of programming ability is quite low, working on them.
If you specifically need to deploy a script as part of a workflow, then Taverna and Galaxy are great. If you actually need a web interface in a more general form, then don't look for something bioinformatics specific.
And if you can produce solid results with PHP, MySQL, and JQuery, just keep going. If you do feel the need to move to one of the newer tools, then look around you and find someone who is already using one of them. I don't mean someone who plays with them, but someone who deploys lots of good code on a regular basis using them and whose work impresses you. Then use what they're using and apprentice yourself to them. If you don't have such a person locally, then don't bother.
For example, we have a guy in my group who does Ruby on Rails. I'm probably a better programmer overall, but for web development, he deploys in that domain faster than I can hope to, and his sites are solid. The biologists love them. Therefore, if I were going to try a framework, I'd use Ruby on Rails and harass him to learn all his tricks.
Related: "What do bioinformaticians use to make high quality web sites?"
I am always wondering: does someone know the distribution of web-dev frameworks among, say, databases published in an NAR issue?