A293561(n) = 0,0,0,3,19,80,286,945...
I independently stumbled on the construction StirlingS2[n + 1, 3] - Binomial[n, 2], which seems to agree for all computable n (I tried up to 100).
The explicit formula is (3^n−2^{n+1}+1−n^2+n)/2
Not sure what the connection is because A293561, which is a column of A142249, only has an analytic definition in terms of polylogarithms.
As for a(n) = StirlingS2[n + 1, 3] - Binomial[n, 2], here is what it counts:
The number of pairs of disjoint subsets of an n-element set that are non-empty and not both singletons.
For n=3:
1|23, 12|3, 13|2
For n=4:
1|23, 1|24, 1|34, 2|13, 2|14, 2|34, 3|12, 3|14, 3|24, 4|12, 4|13, 4|23
1|234, 2|134, 3|124, 4|123,
12|34, 13|24, 14|23,
The StirlingS2[n + 1, 3] gives us the number of partitions of n-set into 3 subsets, 2 of them non-empty. Then Binomial[n, 2] removes the cases where the 2 main subsets only have 1 element each.
This construction arose when I was playing around with
Erdos's Problem 1. Given a set S of n integers, you can obtain all such pairs of disjoint subsets (non-empty, not both singletons), then subtract the sum of one from the sum of the other. If any of the differences are 0, the set is not subset-sum distinct.