Considering the (very!) early ancestors of THEOS are from the Homo-CP/M family... it is very possible that the CP/M compress/uncompress program MAY work, or at least help move you forward.
If you could look at the beginning of the THEOS file, (with a hex-editor) it's very possible to get a good idea of what kind of compression method was used. For the most part, in those days, they used a modified huffman encoding using binary trees. You should be able to see a sort of 'dictionary' near the beginning of the file which defines the tree elements, and the rest of the file is considered as a bit-stream, matching up bit patterns to the dictionary 'words'.
Of course, this is wild conjecture and educated guessing, based on my own experiences with CP/M and it's various hairy descendants.