As is well known, quantized enveloping algebras $U_q(\frak{g})$ admit far fewer subHopf algebras than classical enveloping algebras $U(\frak{g})$. As one can check directly, for appropriate subsets of the (simple root) standard generators $E_i,F_i,K_i$, a Hopf subalgebras are seen to be generated. However, I do not know of subHopf algebras which are not generated by the standard generators. Are there examples of objects? Moreover, since the notion of a Hopf subalgebra is so restrictive for $U_q(\frak{g})$, is it naive to expect some kind of classification?

$\begingroup$ I think there are more simple but "uninteresting" examples than you suggest: how about the Hopf subalgebra generated by $K_1^2$, for example? $\endgroup$– Jan GrabowskiCommented Apr 10, 2018 at 14:43

2$\begingroup$ If you've not done so already, you might also like to look into coideal subalgebras  this is arguably a better notion in this context than Hopf subalgebra. $\endgroup$– Jan GrabowskiCommented Apr 10, 2018 at 14:46

$\begingroup$ Yes, I am aware of coideal subalgebras, but was simply curious as to how badly the process of qdeforming Hopf subalgebras fails. $\endgroup$– Bas WinkelmanCommented Apr 10, 2018 at 15:36

$\begingroup$ I agree it's a reasonable question, for sure. Sorry I couldn't add more than \epsilon in terms of an actual answer... $\endgroup$– Jan GrabowskiCommented Apr 10, 2018 at 15:40
2 Answers
Since we know from EtingofKazhdan that quantization is functorial we can safely say that the classification of subLie bialgebra, which was obtained in the standard case, implies classification of subHopf algebras.
The paper in which classification of Lie bialgebras is obtained is J. Stokman in "the quantum orbit method for generalized flag manifold" arxiv: math/0206245, Proposition 2.1 and basically the result says that yes, you just get subLie bialgebras standardly generated.
Coideal subalgebras, as suggested by Grabowski, are a much richer family; stil some classification results were obtained.
Since the OP is asking for examples of
subHopf algebras which are not generated by the standard generators
i.e. the Chevalley generators (which are actually the generators of the Cartan–Weyl basis with a different normalization), satisfying the Chevalley–Serre Relations, maybe the following method might appear useful for constructing such examples:
The quantum universal enveloping (super)algebras $U_q(g)$, where $g$ is a simple Lie (super)algebra, are known to have alternative—isomorphic—descriptions in terms of $q$deformed paraparticle (parabosonic or parafermionic) algebras: such descriptions are called preoscillator or paraparticle realizations in the mathematical physics literature.
(They extend the corresponding oscillator realizations of the undeformed cases, where UEA of Lie (super)algebras are shown to be isomorphic or homomorphic to the usual (undeformed) parabosonic or parafermionic algebras, or homomorphic to Weyl and Clifford algebras, i.e., the CCR and the CAR relations of quantum mechanics).
See for example Palev  A superalgebra morphism of $U_q(osp(1/2N))$ onto the deformed oscillator superalgebra $W_q(N)$ written in terms of $q$deformed parabosonic generators or Palev  Quantization of $U_q[so(2n + 1)]$ with deformed paraFermi operators written in terms of $q$deformed parafermionic generators.
In the paper Palev and Van der Jeugt  The quantum superalgebra $U_q\big( osp(1/2n)\big)$: deformed paraBose operators and root of unity representations, the authors use deformed realizations to connect the rep theory of $U_q\big( osp(1/2n)\big)$ with the deformed paraboson Fock spaces. They describe the isomorphism between the $q$deformed UEA of the Lie superalgebra $osp(1/2n)$ and the $q$deformed parabosonic algebra: see p. 2608, relations (2.12) and their converse: (2.13), (2.14), (2.15) for the undeformed case and p. 2609, relation (3.3), (3.4) and their converse in p. 2610, rel. (3.5), for the deformed case.
These realizations are then used (see p. 2611, Proposition 4) to provide a description of the subhopf algebra
$$
U_q\big(gl(n)\big)\subset U_q\big(osp(1/2n)\big)
$$
in terms of deformed parabosonic generators. The $U_q\big(gl(n)\big)$ subhopf algebra is described in terms of a suitable subset of deformed parabosonic generators and relations (which are not linear combinations of the CartanWeyl generators).
Similar methods, utilizing bosonic, fermionic or paraparticle generators, can be used to describe—and maybe used as a tool at some classification attempt—other hopf subalgebras of various $q$deformed UEAs. There are lots of works in a similar spirit in the mathematical physics literature—see also the references in the cited articles.
I hope the above might be of some interest for the purposes of the OP.