In the interest of research I investigated the Stanford .ply format v1.0 (exported by blender). It is ascii and fairly straight forward. It required malloc for the vertices and faces and a file read iterative and then display the faces as indexed in the vertex data. It was a few minutes work. The goal is to create a new file format which is .gply that represents the genotype to phenotype expression in a format compatible with the poly-amino-acid control sequence in the cell as well as directional information of association. The individual rule sets of the cellular cpu with its selected function set is the node in a cross-linked tree with a temporal component. What it represents is a linked sequence of cells as vertex centers located at a specific cell and the binary expansion from that point. By including the physical variance and process generation, it also serves as a physics model for animation or simulated movement. Though the system is very complex, the symbolic level of structure is actually very limited. It is not necessary to implement redundant scale to match the bulk properties, except as it influences the physics of its operation. The models are generated by script in or out of blender, and conform to the spatial organization with internal faces removed.
The degree of integration in the program makes it easy to extend for new applications. Memory allocation and freeing is handled by a separate module and display and indicators as well as property selection are also simple interfaces that are flexible.
The program seems to function well, but will require one more complete rewrite to remove vestigial code and look for dangerous corner cases. By using a relative coordinate system, the models become more fluid and extensible. I don't feel that an absolute coordinate model is of much use at all. In order for the system to be extensible in structure and conform to process, each element must have the ability to operate independently within its context , in time as well as mode. A multiple (by time) linked and cross-linked (forward and back) list does represent the structure in such a way that actual genetic switches can be associated with pathways in the model itself.
It is a very versatile and effective system that allows the production of virtually anything of any good use. It is the ultimate RepRap at the molecular level. The programming language is a real monster though. It looks like it was spec'd by random events, Oh right.