My current project involves, in a broad sense, finding ways to fit complex (or mathematically inconvenient) models to data. If you can't fit your models to data, what's the point of the model, or the data?
Specifically, I'm working on statistical inference on protein interaction networks. I'm trying to find ways to fit generative models (such as preferential attachment, duplication-mutation-complementation) to network data in the form of adjacency matrices. Surprisingly little work on the subject has been done so far; a promising approach is Approximate Bayesian Computation.
What's the goal - what are you trying to infer? I don't know what you mean by "preferential attachment, duplication-mutation-complementation".
There must be quite a few undergrad/graduate/post-doc/???-level researchers on LessWrong. I'm interested in hearing about your work. I'll post about myself in the comments.