● What's the phylogenetic bootstrap?: We sample the data, calculate MSA and create a phylogenetic tree. But we need more data to calculate the confidence of our tree. Phylogenetic bootstrap is a trick, that helps us to simulate more someway meaningfull data with the initial MSA, and use it to find the confidence of the original tree.

● How does it work?: Get the distribution of sites of the original MSA, the generatenew fake MSAs with this distribution. Find the trees with maximum likelihood for this fake MSAs. Then compare these trees to the original tree and compute the confidence of the original tree.

● How can we map the bootstrap support to the best-known ML tree?: For each bipartition of the original tree we have to calculate the amount of such bipartitions in the generated fake trees.

● What is a bipartition?: If we have a phylogenetic tree, we can cut it at some brach. This would separate the tree in two bipartitions. A tree is defined by its set of his bipartitions

● What is a trivial bipartition?: bipartitions that split only one sequence from the rest of the tree are called trivial, because any tree generated from this MSA would have this bipartition

● How can we store bipartitions?: the best representation of bipartition is a bit vector. 0 and 1 represents the abscense and presence of a sequence in a substet. They are very memory efficient - 1 bit for one element in bipartition. For the union the bitwise OR can be used, and for the intersection the bitwise AND.

● What is a consensus tree?: that is a tree that summarizes the branching information from all conflicting trees. These trees have a threshold t, that tolerates some amount of conflicts.

● Which flavors of consensus trees do you know?: strict consensus: very conservative, with t near 100%, majority consensus, with t greater than 50%, extended majority TODO

● How can we build a strict/majority rule consensus tree?: we list all bipartitions from the all trees. Compute the consensus tree with bipartitions that ocures in more than t percents of trees. For branch strict constensus trees all trees have to agree with this branch. All bipartitions, that have not enough matches will be smashed together as unresolved part

Notes