Wouldn't it be great if we could have data mining tools that could consider everything there is and build models with which we could have an interesting, even enlightening discussion? Just like talking to a friend or a colleague? I can't say that we are anywhere close. But my lab has in the past years developed some exciting algorithms that can help us digest large volume of data sets, visualize them in a number of fancy ways, and even fuse them together into a single predictive model. For large-scale data fusion, we have lately tested some predictions in the wet labs and achieved some quite astonishing accuracy.
I also enjoy building things. My Bioinformatics Lab develops Orange, a data mining suite with a cool visual programming interface. We are also authors of dictyExpress, a simple gene expression analytics that has found much use within Dictyostelium research community. Our first popular web application was GenePath: it is over ten years old but still runs! And we are teaming up with a spin-off Genialis to build data mining pipelines with simple web interfaces.