Ruiz-Echartea et al present a novel approach for data-driven and multi-docking. Overall, this is a novel real-space docking technique with multiple potential applications in docking problems. This type of research is very novel and under-explored in the community, thus it is certainly worth publishing. Yet, I have several comments to the authors.

General comments:

Could you please comment how the distance constraints can be obtained in practice? Can one, for example, use evolutionary restraints? How well the method is robust against false-positive constraints?

"We showed previously that ... Protein Docking Benchmark v5" - could you please specify where you demonstrated it, will it also imply the same for PPDBv4 and why are you using benchmark v4 in this study?

"to assemble trimers" - Could you please describe the topologies for the trimers and explain how to apply the method to the linear topologies? Specifically, should one know in advance the interaction pattern between the subunits, and how to obtain this pattern in practice and choose the central subunit?

page 5 - do we know what is the pi-ball clash data structure?
"then it is natural to suppose that TBC should be found in the list of B-C pairwise solutions" - here you implicitly assume a triangular topology. Could you please explicitly mention this?

"find a matrix in the B-C list that is similar to T_BC = T_AB ·T_AC−1" These are spatial operators and not matrices!

"to search for such a matrix..." - please see above

"the RMSD is computed between TBC and the transformations composed by the rotation R and the translations ..." - please explain how RMSD is computed or give a reference (in constant time? - this is essential!)

Minor corrections:

• "pair of beads and b of each restraint" - could you please specify that a bead may contain several atoms?

"according the sum" -> "according to the sum"
"HADDOCK using experimental data allows" -> please rephrase
"In such a way that a monomer"
"protein benchmark v4"->"protein-protein docking benchmark v4"
"we therefore let each pair of beads ... correspondS" -> "correspond"
" cone angle thresholds" - extra space before "cone"

page 5 "Figure (a) / (b)" - Figure numbers are missing

caption 1 : "the pair of beads do not satisfy" -> does not. Please also explain what is Rmin.

The authors have addressed all of my previous comments. I have only a few minor remarks left, which are listed below.

Minor comments:

Generally, it's not very convenient (or rather very inconvenient) to not see point-to-point answers to the questions... Please reconsider your responses for the future.

be-tween -> between

con-tact -> contact

"thus speeding up computation (e.g. ClusPro,Eros-DOck)..." - can you please explain what e.g. means? Does it mean that you have these options in ClusPro and Eros-Dock? Maybe it will be clearer to the readers to see smth. like "thus speeding up computation (e.g. as in ClusPro, Eros-DOck)"?

"using experimental data with HADDOCK allows to model symmetric multicomponent assemblies..." - why symmetric? It allows to model nonsymmetric as well, right? Why don't you say "multicomponent, including symmetric"?

"using the Rigid RMSD method in constant time [12]" -> I would say "using the constant-time RigidRMSD method", RigidRMSD without a space. Or "using the RigidRMSD method, computes RMSD in constant time".

"The assembly of the pairwise solutions took between 2 and 8 minutes (6 minutes on average) on 48 CPUs." - A paragraph composed of a single sentence looks strange... Please consider bigger paragraphs.

The authors addressed all of my comments and I believe the manuscript is ready for publication.