The most useful artificial cell model would be one that incorporates information from not just the DNA sequence, but also the transcriptome, proteome, and interactome. Obviously none of these would be complete for any organism at this point, but that's ultimately one of the objectives of this experiment - filling in the gaps. As an example of how this could work, consider the following situation: For yeast, we have the whole genome sequence (equates to most of it in reality), a detailed transcriptome, and a detailed interactome so far. But in all of the papers documenting these data sets, the authors mention that it's not complete. But ideally we would know everything about yeast so we need to find any genes, transcripts, proteins, and interactions we missed. The natural inclination for doing so is to just keep re-doing the work with improved techniques or new insights which would probably work in time, but it would be more efficient to work from what we already have. So then by compiling the `ome data into a single model, missing genes could be predicted by new transcripts and validated by a protein, missing interactions could be predicted by correlation of protein and RNA levels, and so on - importantly all predictions would be in the context of all available information.
There's no question that these predictions could be made without such a model (as has been done thus far), but if everything is together in a single utilizable data set (the ome), then everything can be used at once to validate across all data simultaneously. Although coupling of different `omes in an experiment has been done [2], as far as I know a comprehensive, utilizable resource like The Ome has not been created. Furthermore, it could be accessed remotely from a server and computer cluster that would be capable of running the demanding simulations needed. Simulations could include what happens to everything else when gene A is removed or gene B is added? The output could be a simulation of the resulting genome, transcriptome, proteome, based on the information from the interactome (both genetic and physical).
There's applications for synthetic biology here too since the Ome could be used as a highly-detailed simulation ground for inputing new genes or proteins to see how they should work. Currently, synthetic biologists mostly take `shots in the dark' by simply trying things they may or may not think would work (similar to the approach used to create the first synthetic genome).
Not enough information for your needs? Put some more `omes in as well. More recently groups have characterized the lipidome, metabolome, glycome, and others. Although every `ome added increases the complexity, it also increases the robustness of predictions and the ability to predict things overall.
For the sake of brevity, I've only listed a few ways the Ome could be used, and I used yeast as an example since it is so comprehensively studied so far. Also, the fact that yeast is unicellular makes the situation much simpler since alternate tissue characteristics need not be considered. However, ideally an Ome would be created for many eukaryotic cell types. The unique thing about The Ome is that it doesn't need to be complete to be useful since in a way, it will work towards completing its self if it can be used for predictions. It would also be continually updated from new research findings because individual `omics are still critical.
[1] Synthetic genomes brought closer to life.
Holt RA.
Nat Biotechnol. 2008 Mar;26(3):296-7.
[2] The model organism as a system: integrating 'omics' data sets.
Joyce AR, Palsson BØ.
Nat Rev Mol Cell Biol. 2006 Mar;7(3):198-210.
