I have just made a "Work in Progress" pull request to FCCAnalyses, in which the vertexing code of Franco is now interfaced properly, by linking Delphes:
 
Consequently, one can now use Franco's Vertexmore class. I have added an example that reconstructs the (Ds and) Bs vertices in Bs to Ds(KKpi) K :
It is an improved / cleaner version of the example that I had given to Marco S last summer (which was already using Vertexmore).
 
There are many comments in the example.
It runs over a test file, the corresponding EvtGen card being here:
 
The example is "MC seeded", i.e. I first use the MC truth in order to search for the K, K, K, pi tracks of interest. 
The Ds daughter tracks are fitted to a common vertex; the Ds "pseudo-track" is then reconstructed, and fitted together 
with the track of the bachelor K. The example is run with:

fccanalyses run analysis_Bs2DsK_MCseeded.py --test --nevents 2000
A simple script, plots_Bs2DsK.x in the same directory, makes some basic validation plots.

 

The next step would be to move away from the "MC seeding" and work with all tracks in the event. 
 
I also point you to the SW tutorial and the vertexing examples:
You can see in 2.4.3  how to reconstruct the non-primary tracks (example updated in the PR, following the other changes 
made in this PR), and some stuff in 2.4.5 for the tau -> 3 mu may also be of some interest.