Mapping class groups of Haken Seifert 3-manifolds (not small) This is a somewhat broad question. I would like to get an idea of what is known about mapping class groups (MCG) of a Seifert manifold $M$, in particular how to systematically compute it.
I want to restrict to Haken ones for now, and it is known that $\text{MCG}(M)$ is isomorphic to $\text{Out}(\pi_1(M))$ (c.f. page 33 of https://www.uni-regensburg.de/Fakultaeten/nat_Fak_I/friedl/papers/3-manifold-groups-final-version-031115).
My main focus is $M$ over an $S^2$ base, namely $S^2\left(b;\frac{a_1}{b_1},\dots,\frac{a_n}{b_n}\right)$ (sorry for this non-standard notation). A potentially useful reference is https://arxiv.org/abs/math/0010077 by D. McCullough, whose Table 4 contains only $S^2(2,2,m), S^2(2,3,3), S^2(2,3,4), S^2(2,2,5)$. Their MCGs are known according to Section 3 in the same paper.
I would like to know at least the MCGs of Seifert manifolds $S^2\left(\frac{a_1}{b_1},\frac{a_2}{b_2},\frac{a_3}{b_3}\right)$, which are beyond the scope of the paper by McCullough.
 A: The last case you are interested in is essentially covered by a paper of McCullough and Soma.  For such manifolds (other than three-sphere geometry) the inclusion of the isometry group into the diffeomorphism group is a homotopy equivalence.  Thus the mapping class group is finite, and can be deduced from the symmetries of the base, together with a possible reflection of the fiber. (The assumption of Haken or non-Haken does not play a role.)

When there are more cone points the problem appears to be more delicate.  Hopefully an expert will weight in; here is my understanding of the situation.
Suppose that $M$ is a Seifert fibered space with a unique Seifert fibering.

*

*Thus all mapping classes have a representative that preserves the fibering.  (There is a subtle point here. I think we want the group of all homeomorphisms of $M$ to deformation retract to the subgroup of fibre-preserving homeomorphisms.)

*Drill out a small open fibered neighbourhood $U$ of a generic fibre to obtain $M' = M - U$.

*Thus $M'$ is a surface bundle with fiber $F'$ over the circle.  This gives us a branched covering from $F'$ to the (once-punctured) orbifold base $B'$.  Compute the monodromy of this covering.

*Compute the mapping class group of the base $B'$.

*Compute the finite index subgroup that lifts to give mapping classes of $F'$.

*Each lifted mapping class gives a homeomorphism of $M'$; these act on $\partial M'$.  There is a (index two) subgroup that preserves the meridional slope of $U$.

This feels like an approachable, but definitely non-trivial, problem.
