If you really can figure out how to create an empty host, this should not be too hard. You just need two very important activation genes to come first and then last. Have the first parent fire this gene:
*.dnalen .out1 store
*id# .out2 store
*parent# .out3 store
Now when any species of this kind looks at it they won't attack it and they will know if they should add a gene. Use the out1 to tell the parents which gene to copy. Using the out3 slot will tell which parent to fire their genes into the empty cell. Once out1 = to the parent, the gene copy is completed. Essentially this would occur between one species and hopefully one of them contains some mutation to make a bigger change.
As of now, Genes cannot be split.