Yes, for zero bias, checking forces and total energy convergence is often enough to ensure the reliability of the optimized junction. NEGF provides transport properties, but it's computationally expensive. Geometry optimization with force and energy checks gives a reliable atomic configuration and is usually sufficient for low-bias or zero-bias cases.
This is what i think. For expert view point, please wait for @Dr Blom reply