Skip to main content
. 2019 Jul 5;35(14):i408–i416. doi: 10.1093/bioinformatics/btz312

Fig. 2.

Fig. 2.

An example reduction from the Clique problem to MCT. (a) An undirected graph H with |E(H)|=4 edges and n=|V(H)|=4 vertices, containing a clique of size 3. (b–e) The n =4 input trees T={T1,T2,T3,T4} to the MCT problem obtained from H. The problem instance of determining whether H contains a clique of size c =3 reduces to the MCT instance (T,k) where k=nc+1=2. An optimal clustering σ for (T,2) yields T1={T1,T2,T3} and T2={T4}. (f) The parent-child graph GT1, with the optimal consensus tree R1 for input trees T1 indicated in red. The parent-child graph of T2 is identical to T4 with edge weights (u,v)=1 for each edge (u, v) such that the corresponding optimal consensus tree R2 equals T4. As such, the total distance equals 2[(c1)·|E(H)|(c2)]=10. By Lemma 6, H contains a clique of size c =3