I think Arthur Lander gives the most accessible talks. A lot of his talks around growth control [1] and pattern formation [2] where he shows the tradeoffs between different cellular network and feedback mechanisms and how this influences the robustness of development. Then you have people like me [3] and my PhD adviser Qing Nie [4] where we describe how noise, the randomness that exists in biochemical interactions, is able to be controlled and used by biological organisms to improve developmental processes. And then there's people like John Lowengrub who talk about the morphology of tumors [5] to be able to predict and describe how the behavior can be derived from the shape.
There's a lot of movement going on in the data generation side of systems biology. Single-cell RNA-seq methods allow us to quantify almost all of the RNA generated by each individual cell for a whole group of cells, which lets us know exactly how cells are generating different proteins and by what amounts [6]. This of course opens up a whole lot of questions, along the big data computing side (how do you handle such a large dataset and extract something new out of?) and from the mathematical side (what kind of models is this actually validating and invalidating?). With this kind of data, there are questions about how well we can fit computational models directly to data and improve prediction accuracy beyond qualitative results [7]. These models are becoming precise enough that people are starting to test their uses in pharmocology, doing personalized drug-dosage prediction using covariates and large internal systems biological models called PBPK models [8].
[1]: https://www.youtube.com/watch?v=Dza0hM2vBNk
[2]: https://www.youtube.com/watch?v=8jFx9I7lBOg
[3]: https://www.youtube.com/watch?v=_h5fVDvGp-8
[4]: https://www.youtube.com/watch?v=tRzrBvGqOOE
[5]: https://www.youtube.com/watch?v=NVFFoGfC4Uk
[6]: https://www.youtube.com/watch?v=8KpIDpPEGV4
