Okay. Imagine I give you an organisms DNA. You don’t know what organism it is, under what conditions it lives, or anything. I just give you it’s DNA sequence and nothing else. Now, you can’t cheat and go compare it’s genome to other organisms in some database to get some sort of idea of how it lives and behaves and what it’s phenotypes are likely to be (if you do, you’re getting your information from something other than DNA).
Can you predict from that alone the physiology and behavior of the organism? If I were to provide you a hypothetical environment (say it lives at 90 degrees C water in some geothermal hotspring), can you predict it’s expression profile from that at any given moment?
No, you can’t do that because you have no idea what genes will be expressed at that moment. Not even with a supercomputer to sort of simulate it, because all the simulation has to work with is a piece of DNA. That does not an organism make.
The reason you don’t know what genes will be expressed is that I haven’t told you the information you need to predict that. That is information about the cytoplasmic contents. What does the cytoplasm (including the nucleus and other organelles if it’s a eukaryote) contain? What sits in the membrane? Is it a double membrane? What is the composition of that membrane? Is there a cell wall? Is it actively being constructed or degraded?
Regulatory elements, metabolites, signaling factors, transport proteins, membrane channels, receptors etc.
It’s expression profile, and subsequent behavior going forward, significantly depends on what proteins, metabolites, signaling factors, and other regulatory molecules were previously expressed. If there is dextrose in the environment, will the organism take that in? Is there a receptor or transport protein for that? Is it active?
Since you don’t have this information you can’t predict it’s behavior. If you can’t predict it’s behavior you’re missing information. So the cytoplasm and membrane and all that other stuff must contain information too.
If I were to give you the full set of information about the state of the organism at some instant (what it contains, where these contents are located in relation to each other, what it’s membrane is like and so on), then you can predict it’s behavior (at least with a sufficiently powerful computer) going forward. You couldn’t possibly do this with the DNA alone.
That’s really all there is to it.
DNA alone has no “agency”. DNA is replicated by the interaction of many constituents of the cell, most of which themselves are the products of the transcription and translation of DNA, and this mutual dependence and cycle of replication, transcription, translation, and so on must go back to the origin of cells, and if genes came before cells, back to even that.