Having worked in the field of protein recruitment simulations, I can confirm this. We need to know a lot more before we can actually build a simulation of the entire protein dynamics in a cell (which is what would be needed to make any sort of reliable sim-model-only predictions).

Computer simulations _do_ help in guiding the research by showing gaps in the theoretical understanding and identifying candidate proteins/genes for further experimental investigation, but no researcher in their right mind would take some finding from a current cell-biological computer simulation and publically declare "We've identified the cause for cancer X".