Epigenetics are still structure that could theoretically be replicated.

Talk of replication is hypothetical: we're very far from that level of precise control. It's not theoretically impossible, though, to have something that's a functional replica down to the level of individual proteins. The same is true for neural impulses: no matter how subtle and sublime they may be, they're ultimately chemical/electrical signals that could be precisely replicated with suitably advanced technology. For a brain in a vat, there is no difference between a real touch from a lover and the simulated equivalent, so long as all input is the same.

We can't say whether a 'replicant' (for lack of a better term) would be conscious, but we're also fundamentally unable to demonstrate that other humans are conscious, beyond asking them and trusting their responses.

The replicant wouldn't be devoid of attachment and interpersonal connection, either. If we're replicating the environmental inputs, that would all be part of the simulation. Supposing we can do all that, and that a brain thinks it has lived a normal life and had a normal childhood, why should we expect different outputs because the environment is simulated and not based on input from organic sensory organs?