In this recent article "Design and synthesis of a minimal bacterial genome", they created a minimal cell with only 473 genes. However, they didn't synthesize all of the necessary components of the cell to get going. How much of those extra stuff can we currently synthesize, and what are the ones that we can't synthesize in the lab to create a working cell from scratch?
Designing/engineering functional proteins from scratch is feasible but would take a ridiculous amount of work. Michael Hecht is one person I know of who has done this (example: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0015364). J Craig Venter wants to determine the minimal set of molecular machines required for life, regardless of their origin. This would begin to illuminate which protein functions are absolutely required for life. I believe we could design and engineer proteins de novo to perform all of those functions in theory, but in practice it would be unrealistic. I think the key point to glean from their synthetic cell is the set of functions, not specific genes, required for life. Those functions could be performed by evolved or synthetic molecules.
As a side note, while I believe that to be the overarching goal, they do note the surprising lack of functional characterization in the syn3.0 genome.