 Research
 Open Access
 Published:
Triplet supertree heuristics for the tree of life
BMC Bioinformatics volume 10, Article number: S8 (2009)
Abstract
Background
There is much interest in developing fast and accurate supertree methods to infer the tree of life. Supertree methods combine smaller input trees with overlapping sets of taxa to make a comprehensive phylogenetic tree that contains all of the taxa in the input trees. The intrinsically hard triplet supertree problem takes a collection of input species trees and seeks a species tree (supertree) that maximizes the number of triplet subtrees that it shares with the input trees. However, the utility of this supertree problem has been limited by a lack of efficient and effective heuristics.
Results
We introduce fast hillclimbing heuristics for the triplet supertree problem that perform a stepwise search of the tree space, where each step is guided by an exact solution to an instance of a local search problem. To realize time efficient heuristics we designed the first nontrivial algorithms for two standard search problems, which greatly improve on the time complexity to the best known (naïve) solutions by a factor of n and n^{2} (the number of taxa in the supertree). These algorithms enable largescale supertree analyses based on the triplet supertree problem that were previously not possible. We implemented hillclimbing heuristics that are based on our new algorithms, and in analyses of two published supertree data sets, we demonstrate that our new heuristics outperform other standard supertree methods in maximizing the number of triplets shared with the input trees.
Conclusion
With our new heuristics, the triplet supertree problem is now computationally more tractable for largescale supertree analyses, and it provides a potentially more accurate alternative to existing supertree methods.
Background
Assembling the tree of life, or the phylogeny of all species, is one of the grand challenges in evolutionary biology. Supertree methods take a collection of species trees with overlapping, but not identical, sets of taxa and return a "supertree" that contains all taxa found in the input trees (e.g., [1–4]). Thus, supertrees provide a way to synthesize small trees into a comprehensive phylogeny representing large sections of the tree of life. Recent supertree analyses have produced the first complete familylevel phylogeny of flowering plants [5], and the first phylogeny of nearly all extant mammals [6]. Since the main objective of most supertree analyses is to build extremely large phylogenetic trees by solving intrinsically hard computational problems, the design of efficient and effective heuristics is a critically important part of developing any useful supertree method.
Ideal supertree methods must combine speed and accuracy. By far the most commonly used supertree method is matrix representation with parsimony (MRP; [7, 8]). MRP converts a collection of input trees into a binary character matrix, and then performs a parsimony analysis on a matrix representation of the input trees. Thus, MRP analyses can use efficient parsimony heuristics implemented in programs such as PAUP* [9] and TNT [10], making largescale MRP supertree analyses computationally more tractable. However, the accuracy and performance of MRP are frequently criticized. For example, there is evidence of input tree size and shape biases [11, 12], the results can vary depending on the method of matrix representation [11], and the accuracy of the MRP supertrees are not necessarily correlated with the parsimony score [13]. Therefore, there is a need to develop alternate methods that share the advantages of MRP but produce more accurate supertrees.
Since we rarely know the evolutionary history of a group of organisms with certainty, it is usually impossible to assess the accuracy of a supertree based on its similarity to the true species phylogeny. A more practical way to define the accuracy of a supertree is based on the overall similarity of the supertree to the collection of input trees. There are numerous ways to measure the similarity between input trees and the supertree. The intrinsically hard [14] triplet supertree problem measures this similarity based on the common shared triplets, or rooted, binary, 3taxon trees that are the irreducible unit of phylogenetic information in rooted trees [14]. Specifically, the triplet supertree problem seeks a supertree that shares the most triplets with the input trees.
We introduce hillclimbing heuristics for the triplet supertree problem that make it feasible for truly largescale phylogenetic analyses. Hillclimbing heuristics have been effectively applied to other intrinsically difficult supertree problems [7, 13, 15]. They search the space of all possible supertrees guided by a series of exact solutions to instances of a local search problem. The local search problem is to find an optimal phylogenetic tree that shares the most number of triplets with the input trees in the neighborhood of a given tree. The neighborhood is the set of all phylogenetic trees into which the given tree can be transformed by applying a tree edit operation. A variety of different tree edit operations have been proposed [16, 17], and two of them, rooted Subtree Pruning and Regrafting (SPR) and Tree Bisection and Reconnection (TBR), have shown much promise for phylogenetic studies [18, 19]. However, algorithms for local search problems based on SPR and TBR operations, especially on rooted trees, are still in their infancy. To conduct largescale phylogenetic analyses, there is much need for effective SPR and TBR based local search problems that can be solved efficiently.
In this work we improve upon the best known (naïve) solutions for the SPR and TBR local search problems by a factor of n and n^{2} (the number of taxa in the supertree) respectively. This is especially desirable since standard local search heuristics for the triplet supertree problem typically involve solving several thousand instances of the local search problem. We demonstrate the performance of our new triplet heuristics in a comparative analysis with other standard supertree methods.
Related work
Triplet supertree problem
The triplet supertree problem makes use of the fact that every rooted tree can be equivalently represented by a set of triplet trees [17]. A triplet tree is a rooted fully binary tree over three taxa. Thus, a tripletsimilarity measure can be defined between two rooted trees that is the cardinality of the intersection of their triplet presentations. This measure can be extended to measure the similarity from a collection of rooted input trees to a rooted supertree, by summing up the tripletsimilarities for each input tree and the supertree. The triplet supertree problem is to find a supertree that maximizes the tripletsimilarity for a given collection of input trees. Figure 1 illustrates the triplet supertree problem.
Hillclimbing heuristics
We introduce hillclimbing heuristics to solve the triplet supertree problem. Hillclimbing heuristics have been successfully applied to several intrinsically complex supertree problems. In these heuristics a tree graph is defined for the given set of input trees and some, typically symmetric, treeedit operation. The nodes in the tree graph are the phylogenetic trees over the overall taxon set of the input trees. An edge adjoins two nodes exactly if the corresponding trees can be transformed into each other by the tree edit operation. The cost of a node in the graph is the measurement from the input trees to the tree represented by the node under the particular supertree problems optimization measurement. For the triplet supertree problem, the cost of a node in the graph is the tripletsimilarity from the input trees to the tree represented by the node. Given a starting node in the tree graph, the heuristic's task is to find a maximallength path of steepest ascent in the cost of its nodes and to return the last node on such a path. This path is found by solving the local search problem for every node along the path. The local search problem is to find a node with the maximum cost in the neighborhood (all adjacent nodes) of a given node. The neighborhood searched depends on the edit operation. Edit operations of interest are SPR and TBR [17]. We defer the definition of these operations to the next section. The best known run times (naïve solutions) for the SPR and TBR based local search problems under the tripletsimilarity measurement are O(kn^{4}) and O(kn^{5}) respectively, where k is the number of input gene trees and n is the number of taxa present in the input gene trees.
Contribution of the manuscript
We introduce algorithms that solve the local SPR and TBR based search problems for our triplet supertree heuristics in times O(n^{3}) and O(n^{3}) respectively, with an initial preprocessing time of O(kn^{3}). These algorithms allow true largescale phylogenetic analyses using hillclimbing heuristics for the triplet supertree problem. Finally, we demonstrate the performance of our SPR and TBR based hillclimbing heuristics in comparative studies on two large published data sets.
Methods
Initially, for each possible triplet over the set of all taxa we count and store the frequency displayed by all the input trees in O(kn^{3}) time. Then, for each local search problem, we use dynamic programming to efficiently preprocess necessary triplet counts in O(n^{3}) time. By exploiting the structural properties of SPR and TBR related to tripletsimilarity, we are able to use these triplet counts to compute the differences in tripletsimilarity for all SPR and TBR neighborhoods, each in O(n^{3}) time.
Basic definitions, notations, and preliminaries
In this section we introduce basic definitions and notations and then define preliminaries required for this work. For brevity the proofs of Lemmas 2–6 are omitted, but available on request.
Basic definitions and notations
A tree T is a connected graph with no cycles, consisting of a node set V(T) and an edge set E(T). T is rooted if it has exactly one distinguished node called the root which we denote by Ro(T).
Let T be a rooted tree. We define ≤_{ T }to be the partial order on V(T) where x ≤_{ T }y if y is a node on the path between Ro(T) and x. If x ≤_{ T }y we call x a descendant of y, and y an ancestor of x. We also define x <_{ T }y if x ≤_{ T }y and x ≠ y, in this case we call x a proper descendant of y, and y a proper ancestor of x.
The set of minima under ≤_{ T }is denoted by Le(T) and its elements are called leaves. If {x, y} ∈ E(T) and x ≤_{ T }y then we call y the parent of x denoted by Pa_{ T }(x) and we call x a child of y. The set of all children of y is denoted by Ch_{ T }(y). If two nodes in T have the same parent, they are called siblings. The least common ancestor of a nonempty subset L ⊆ V(T), denoted as lca_{ T }(L), is the unique smallest upper bound of L under ≤_{ T }.
If e ∈ E(T), we define T/e to be the tree obtained from T by identifying the ends of e and then deleting e. T/e is said to be obtained from T by contracting e. If v is a vertex of T with degree one or two, and e is an edge incident with v, the tree T/e is said to be obtained from T by suppressing v.
The restricted subtree of T induced by a nonempty subset L ⊆ V(T), denoted as TL, is the tree induced by L where all internal nodes with degree two are suppressed, with the exception of the root node. The subtree of T rooted at node y ∈ V(T), denoted as T_{ y }, is the restricted subtree induced by {x ∈ V(T): x ≤_{ T }y}.
T is fully binary if every node has either zero or two children. Throughout this paper, the term tree refers to a rooted fully binary tree.
The triplet supertree problem
We now introduce necessary definitions to state the triplet supertree problem. A triplet is a rooted binary tree with three leaves. A triplet T with leaves a, b, and c is denoted abc if lca_{ T }({a, b}) is a proper descendant of the root. Note that we do not distinguish between abc and bac. The set of all triplets of a tree T, denoted as Tr(T), is {abc : T{a, b, c} = abc}. The set of common triplets between two trees T_{1} and T_{2}, denoted as S(T_{1}, T_{2}), is Tr(T_{1}) ∩ Tr(T_{2}). A profile P is a tuple of trees (T_{1},..., T_{ n }), we extend the definition of leaf set to profiles as $\text{Le}(P)={\displaystyle {\cup}_{i=1}^{n}\text{Le}({T}_{i})}$. Let P be a profile, we call T* a supertree of P if Le(T*) = Le(P).
We are now ready to define the triplet supertree problem (Fig. 1).
Definition 1 (Triplet similarity). Given a profile P = (T_{1},..., T_{ n }) and a supertree T* of P, we define the tripletsimilarity score $\mathcal{S}(P,{T}^{\ast})={\displaystyle {\sum}_{i=1}^{n}S({T}_{i},{T}^{\ast})}$.
Problem 1 (The triplet supertree problem). Given a profile P, find a supertree T* that maximizes S(P, T*). We call any such T* a triplet supertree.
Theorem 1 ([14]). The triplet supertree problem is NPhard.
Local search problems
Here we first provide definitions for the reroot (RR), TBR, and SPR edit operations and then formulate the related local search problems. Figures 2 and 3 illustrate the RR and TBR edit operations respectively.
Definition 2 (RR operation). Let T be a tree and x ∈ V(T). RR_{ T }(x) is defined to be the tree T if x = Ro(T). Otherwise, RR_{ T }(x) is the tree that is obtained from T by (i) suppressing Ro(T), and (ii) subdividing the edge {Pa_{ T }(x), x} by a new root node. We define the following extension:
RR_{ T }= ∪_{x∈V(T)}{RR_{ T }(x)}.
Let x ≤_{ T }v, we also define a partial RR operation RR_{ T }(v, x) by replacing T_{ v }with ${\text{RR}}_{{T}_{v}}$(x).
Definition 3 (TBR operation). For technical reasons we first define for a tree T the planted tree Pl(T) that is the tree obtained by adding an additional edge, called root edge, {r, Ro(T)} to E(T).
Let T be a tree, e = (u, v) ∈ E(T), and X, Y be the connected components that are obtained by removing edge e from T where v ∈ X and u ∈ Y. We define TBR_{ T }(v, x, y) for x ∈ X and y ∈ Y to be the tree that is obtained from Pl(T) by first removing edge e, then replacing the component X by RR_{ X }(x), and then adjoining a new edge f between x' = Ro(RR_{ X }(x)) and Y as follows:
1. Create a new node y' that subdivides the edge (Pa_{ T }(y), y).
2. Adjoin the edge f between nodes x' and y'.
3. Suppress the node u, and rename x' as v and y' as u.
4. Contract the root edge.
We say that the tree TBR_{ T }(v, x, y) is obtained from T by a tree bisection and reconnection (TBR) operation that bisects the tree T into the components X, Y and reconnects them above the nodes x, y.
We define the following extensions for the TBR operation:
1. TBR_{ T }(v, x) = ∪_{y∈Y}TBR_{ T }(v, x, y)
2. TBR_{ T }(v) = ∪_{x∈X}TBR_{ T }(v, x)
3. TBR_{ T }= ∪_{(u, v)∈E(T)}TBR_{ T }(v)
An SPR operation for a tree T can be briefly described through the following steps: (i) prune some subtree S from T, (ii) add a root edge to the remaining tree T', (iii) regraft S into an edge of the remaining tree T', (iv) contract the root edge. For our purposes we define the SPR operation as a special case of the TBR operation.
Definition 4 (SPR operation). Let T be a tree, e = (u, v) ∈ E(T), and X, Y be the connected components that are obtained by removing edge e from T where v ∈ X and u ∈ Y. We define SPR_{ T }(v, y) for y ∈ Y to be TBR_{ T }(v, v, y). We say that the tree SPR_{ T }(v, y) is obtained from T by a subtree prune and regraft (SPR) operation that prunes subtree T_{ v }and regrafts it above node y.
We define the following extensions of the SPR operation:
1. SPR_{ T }(v) = ∪_{y∈Y}SPR_{ T }(v, y)
2. SPR_{ T }= ∪_{(u, v)∈E(T)}SPR_{ T }(v)
Problem 2 (TBR Scoring (TBRS)). Given a profile P and a supertree T of P, find a tree T* ∈ TBR_{ T }such that $\mathcal{S}(P,{T}^{\ast})=ma{x}_{{T}^{\prime}\in {\text{TBR}}_{T}}\mathcal{S}(P,{T}^{\prime})$
Problem 3 (TBRRestricted Scoring (TBRRS)). Given a profile P, a supertree T of P, and (u, v) ∈ E(T), find a tree T* ∈ TBR_{ T }(v) such that $\mathcal{S}(P,{T}^{\ast})=ma{x}_{{T}^{\prime}\in {\text{TBR}}_{T}}\mathcal{S}(P,{T}^{\prime})$
The problems SPR Scoring (SPRS) and SPRRestricted Scoring (SPRRS) are defined analogously to the problems TBRS and TBRRS respectively.
Further, we observe that to solve any of these four local search problems, it is sufficient to find a tree within the neighborhood that gives the maximum increase on $\mathcal{S}$(P, T), without calculating the value of each $\mathcal{S}$(P, T) itself. With this observation, it is useful to give the following definition.
Definition 5. Let P be a profile, T_{1} and T_{2} be two supertrees of P, we define the score difference function, denoted as Δ_{ P }(T_{1}, T_{2}), to be $\mathcal{S}$(P, T_{2})  $\mathcal{S}$(P, T_{1}).
Solving the SPRRS and SPRS problems
We first show how to solve the SPRRS problem. Extending on this solution we introduce a new algorithm for the SPRS problem.
Solving the SPRRS problem
Given a profile P, a supertree T of P, and (u, v) ∈ E(T), we compute Δ_{ P }(T, T') for each T' ∈ SPR_{ T }(v) by first pruning and regrafting T_{ v }to Ro(T) and compute the score differences for each "movedown" operation, then traverse T in preorder to obtain the tree that gives the maximum score difference. We first give a definition that helps us describe a single "movedown".
Definition 6 (Immediate Triplet). Let T be a tree and v ∈ V(T), an immediate triplet induced by v, denoted as yz ▷◁ v, is a triplet yzv where there exists nodes a, b ∈ V(T) such that Pa_{ T }(y) = Pa_{ T }(z) = b and Pa_{ T }(b) = Pa_{ T }(v) = a.
Algorithm 1 Algorithm for the SPRRS problem
1: procedure SPRRS(P, T, (u, v))
Input: A profile P = (T_{1},..., T_{ n }), a supertree T of P, and (u, v) ∈ E(T)
Output: T* ∈ SPR_{ T }(v), and Δ_{ P }(T, T*)
2: r ← Ro(T)
3: $\widehat{T}$ ← SPR_{ T }(v, r)
4: Call MovedownAndCompute(P, $\widehat{T}$, v)
5: Traverse the tree T_{ r }in preorder to compute Δ_{ P }($\widehat{T}$, T') for each T' ∈ SPR_{ T }(v) using the values computed by MovedownAndCompute
6: T* ← T' ∈ SPR_{ T }(v) such that ${\Delta}_{P}(T,{T}^{\prime})=ma{x}_{{T}^{\u2033}\in {\text{SPR}}_{T}}{\Delta}_{P}(\widehat{T},{T}^{\u2033})$
7: d ← Δ_{ P }($\widehat{T}$, T*)  Δ_{ P }($\widehat{T}$, T)
8: return (T*, d)
9: end procedure
10: procedure MOVEDOWN AND COMPUTE(P, T, v)
Input: A profile P, a tree T, and v ∈ V(T)
11: yz ▷◁ v ← The immediate triplet induced by v in T
12: for all t ∈ {y, z} do
13: T' ← SPR_{ T }(v, t)
14: Compute and store Δ_{ P }(T, T')
15: Call MovedownAndCompute(P, T', v)
16: end for
17: end procedure
It can be easily seen that Algorithm 1 is correctly solving the SPRRS problem.
Solving the SPRS problem
Algorithm 2 Algorithm for the SPRS problem
1: procedure SPRS(P, T)
Input: A profile P = (T_{1},..., T_{ n }), a supertree T of P
Output: T* ∈ SPR_{ T }, and Δ_{ P }(T, T*)
2: for all (u, v) ∈ E(T) do
3: Store the value of SPRRS(P, T, (u, v))
4: end for
5: (T*, d) ← the stored value of SPRRS calls that has the maximum score increase by traversing the tree T in postorder
6: return (T*, d)
7: end procedure
Algorithm 2 gives a trivial extension of Algorithm 1 to solve the SPRS problem.
Computing Δ_{ P }(T, T') efficiently
Algorithm 1 assumed the computation of Δ_{ P }(T, T') for each movedown operation (Line 14). In this section we show how to compute each Δ_{ P }(T, T') efficiently by exploiting structural properties related to the tripletsimilarity. We begin with some useful definitions.
Definition 7. Let A, B, C be pairwise mutual exclusive leaf sets, we extend the triplet notation by ABC = {abc : a ∈ A, b ∈ B, c ∈ C}. Further, let u, v, w be three nodes in a tree T having no ancestral relationships, define uv_{ T }w = Le(T_{ u }) Le(T_{ v }) Le(T_{ w })
Definition 8. The Boolean value of a statement ϕ, denoted as ⟦ϕ⟧, is 1 if ϕ is true, 0 otherwise.
Definition 9. Given a profile P = (T_{1},..., T_{ n }) and distinct a, b, c ∈ Le(P), we define the triplet summation function by
Let A, B, C ⊆ Le(P) be pairwise mutual exclusive leaf sets, we extend the triplet summation function by
Further, let u, v, w be three nodes in a tree T having no ancestral relationships, we define
σ_{P, T}(uvw) = σ_{ P }(uv_{ T }w)
Lemma 1. Let T be a tree and yz ▷◁ v be an immediate triplet induced by a node v ∈ V(T). If T' = SPR_{ T }(v, y), then
Tr(T') = (Tr(T)\yz_{ T }v) ∪ vy_{ T }z

(1)
Proof. Let a, b, c ∈ Le(T), we consider the following cases:

1.
If any one of a, b, c is not in the subtree ${T}_{{\text{Pa}}_{T}(v)}$, then T'{a, b, c} = T{a, b, c}. Since both yz_{ T }v and vy_{ T }z only contain triplets formed under the subtree ${T}_{{\text{Pa}}_{T}(v)}$, T{a, b, c} and only this triplet resolution of {a, b, c} is in both sides of equality.

2.
Consider three subtrees T_{ v }, T_{ y }, T_{ z }. If a, b, c are all in one of the subtree, then since the subtrees T_{ v }, T_{ y }, T_{ z }do not change by the SPR operation, T'{a, b, c} = T{a, b, c}. Since both yz_{ T }v and vy_{ T }z only contain triplets that are formed by one leaf from each of T_{ v }, T_{ y }, and T_{ z }subtrees, T{a, b, c} and only this triplet resolution of {a, b, c} is in both sides of equality.

3.
Consider three subtrees T_{ v }, T_{ y }, T_{ z }. If two leaves of {a, b, c} are in one subtree and the other leaf is in another subtree, and suppose WLOG that {a, b} are in one subtree, then we observe that lca_{ T }({a, b}) <_{ T }lca_{ T }({a, b, c}) and lca_{ T' }({a, b}) <_{ T' }lca_{ T' }({a, b, c}), so T'{a, b, c} = T{a, b, c} = abc.
Also, as in Case 2, both yz_{ T }v and vy_{ T }z does not contain triplet formed by {a, b, c}, so T{a, b, c} and only this triplet resolution of {a, b, c} is in both sides of equality.

4.
If each of T_{ v }, T_{ y }, and T_{ z }contains exactly one leaf in {a, b, c}, and suppose WLOG that a ∈ Le(T_{ v }), b ∈ Le(T_{ y }), and c ∈ Le(T_{ z }). Then T{a, b, c} = bca and T'{a, b, c} = abc. Also we observe that bca ∈ yz_{ T }v and abc ∈ vy_{ T }z, therefore RHS, and hence both sides contain abc and only this resolution of {a, b, c}.
Lemma 2. Given a profile P and a supertree T of P, let yz ▷◁ v be an immediate triplet induced by a node v ∈ V(T). If T' = SPR_{ T }(v, y), then
Δ_{ P }(T, T') = σ_{P, T}(vyz)  σ_{P, T}(yzv)
Lemma 3. Given a profile P and a supertree T of P, let distinct v, b ∈ V(T) such that v ≰_{ T }b, b ≰_{ T }v, and Ch_{ T }(b) = {y, z}. If T_{1} = SPR_{ T }(v, b) and T_{2} = SPR_{ T }(v, y), then
Δ_{ P }(T_{1}, T_{2}) = σ_{P, T}(vyz)  σ_{P, T}(yzv)
Lemma 4. Given a profile P and a supertree T of P, let v, b ∈ V(T) such that Pa_{ T }(Pa_{ T }(v)) ≤_{ T }b, and let ∏ = {x ∈ V(T): v <_{ T }x <_{ T }b}. Define C = ∪_{x∈∏}{c∈Ch_{ T }(x):v≰_{T}c}. Let Ch_{ T }(b) = {y, z}, and WLOG suppose y ∉ ∏. If T_{1} = SPR_{ T }(v, b) and T_{2} = SPR_{ T }(v, y), then
Equations (3) and (4) provide computations for all SPRRS neighborhoods (Line 14 of Algorithm 1). We now show how to compute them efficiently.
Algorithm 3 Algorithm to compute triplet summation function
1: procedure PREPROCESS TRIPLET SUM(P)
Input: A profile P = (T_{1},..., T_{ n })
2: Initialize all values of σ_{ P }to 0
3: for i = 1 to n do
4: for all u ∈ V(T_{ i }) in postorder, u ∉ Le(T_{ i }) do
5: {v, w} ← Ch_{ T }(u)
6: for all {x, y} ∈ Le(T_{ v }), z ∈ Le(T_{ w }) do
7: Increment σ_{ P }(xyz)
8: end for
9: for all {x, y} ∈ Le(T_{ w }), z ∈ Le(T_{ v }) do
10: Increment σ_{ P }(xyz)
11: end for
12: end for
13: end for
14: end procedure
Given a profile P = (T_{1},..., T_{ n }), we start by computing the triplet summation function σ_{ P }for all triplets, as shown by Algorithm 3.
Algorithm 4 Algorithm to compute extended triplet summation function
1: procedure PREPROCESS EXTENDED TRIPLET SUM(P, T)
Input: A profile P = (T_{1},..., T_{ n }), a supertree T of P
2: for all u ∈ V(T) in postorder do
3: for all v ∈ V(T) in postorder after u, lca_{ T }({u, v}) ∉ {u, v} do
4: for all w ∈ V(T) in postorder after v, lca_{ T }({u, v, w}) ∉ {u, v, w} do
5: if w ∈ Le(T) then
6: if v ∈ Le(T) then
7: if u ∈ Le(T) then
8: σ_{P, T}(uvw) ← σ_{ P }(uvw)
9: σ_{P, T}(uwv) ← σ_{ P }(uwv)
10: σ_{P, T}(vwu) ← σ_{ P }(vwu)
11: else
12: {u_{1}, u_{2}} ← Ch_{ T }(u)
13: σ_{P, T}(uvw) ← σ_{P, T}(u_{1}vw) + σ_{P, T}(u_{2}vw)
14: σ_{P, T}(uwv) ← σ_{P, T}(u_{1}wv) + σ_{P, T}(u_{2}wv)
15: σ_{P, T}(vwu) ← σ_{P, T}(vwu_{1}) + σ_{P, T}(vwu_{2})
16: end if
17: else
18: {v_{1}, v_{2}} ← Ch_{ T }(v)
19: σ_{P, T}(uvw) ← σ_{P, T}(uv_{1}w) + σ_{P, T}(uv_{2}w)
20: σ_{P, T}(uwv) ← σ_{P, T}(uwv_{1}) + σ_{P, T}(uwv_{2})
21: σ_{P, T}(vwu) ← σ_{P, T}(v_{1}wu) + σ_{P, T}(v_{2}wu)
22: end if
23: else
24: {w_{1}, w_{2}} ← Ch_{ T }(w)
25: σ_{P, T}(uvw) ← σ_{P, T}(uvw_{1}) + σ_{P, T}(uvw_{2})
26: σ_{P, T}(uwv) ← σ_{P, T}(uw_{1}v) + σ_{P, T}(uw_{2}v)
27: σ_{P, T}(vwu) ← σ_{P, T}(vw_{1}u) + σ_{P, T}(vw_{2}u)
28: end if
29: end for
30: end for
31: end for
32: end procedure
Next, we compute the extended triplet summation function σ_{P, T}, for all nodes u, v, w in T having no ancestral relationships, as shown by Algorithm 4.
Algorithm 5 Algorithm to compute score difference function
1: procedure PREPROCESS SCORE DIFFERENCE(P, T)
Input: A profile P = (T_{1},..., T_{ n }), a supertree T of P
2: for all v ∈ V(T) do
3: for all (b, y) ∈ E(T): b ≰_{ T }Pa_{ T }(v) do
4: Let Ch_{ T }(b) = {y, z}
5: Let T_{1} = SPR_{ T }(v, b), T_{2} = SPR_{ T }(v, y)
6: if v ≤_{ T }b then
7: for all p ∈ V(T): v <_{ T }p <_{ T }b do
8: Let x ∈ Ch_{ T }(p) where v ≰_{ T }x
9: Δ_{ P }(T_{1}, T_{2}) ← Δ_{ P }(T_{1}, T_{2}) + σ_{P, T}(vyx)  σ_{P, T}(yxv)
10: end for
11: else
12: Δ_{ P }(T_{1}, T_{2}) ← σ_{P, T}(vyz)  σ_{P, T}(yzv)
13: end if
14: end for
15: end for
16: end procedure
Finally, we compute the score difference function Δ_{ P }(T_{1}, T_{2}) for all v ∈ V(T) and (b, y) ∈ E(T) where b ≰_{ T }Pa_{ T }(v) such that T_{1} = SPR_{ T }(v, b) and T_{2} = SPR_{ T }(v, y). This is shown by Algorithm 5.
Time complexity
We describe the time complexity for our solution for the SPRS problem. First, we run Algorithm 3 once for the entire heuristic run, which takes O(kn^{3}) where k is the number of input trees and n is the number of taxa present in the input trees. Then, for each SPRS problem, we begin by preprocessing necessary counts using Algorithm 4 and 5, each takes time O(n^{3}). These preprocessed computations allow Algorithm 1 to run in O(n) time. Finally, Algorithm 2 issues O(n) calls to Algorithm 1, so overall it takes O(n^{2}) time. Including the preprocessing steps, we see that solving the SPRS problem takes O(n^{3}) time, with an expense of O(kn^{3}) at the beginning of the entire heuristic run.
Solving the TBRRS and TBRS problems
We extend our solutions of SPRRS and SPRS problems to solve TBRRS and TBRS problems.
Solving the TBRRS problem
We observe that a TBR operation can be viewed as an SPR operation followed by an RR operation. We exploit this structural property further and establish some lemmas which helps us compute the score differences for all TBR operations in the TBRRS neighborhood.
Lemma 5. Given a profile P, a supertree T of P, and a valid TBR operation on T where T' = TBR_{ T }(v, x, y), then
Δ_{ P }(T, T') = Δ_{ P }(T, SPR_{ T }(v, y)) + Δ_{ P }(T, RR_{ T }(v, x))
Lemma 5 implies that given a subtree, we can find the best TBRRS neighborhood by finding the best SPRRS neighborhood and apply the best rerooting for the subtree regardless of which SPR operation was chosen. Further, we note that RR is a special case of SPR operation by the following lemma.
Lemma 6. Given a profile P and a supertree T of P, let xy ▷◁ z be an immediate triplet induced by a node z ∈ V(T) and Pa_{ T }(z) = v, then
RR_{ T }(v, x) = SPR_{ T }(z, y)
This means that we can reuse Lemmas 2, 3, 4 and their corresponding algorithms to compute the score differences of all movedowns in terms of rerooting. Hence, given a subtree the algorithm would first compute the best rerooting and its score difference, then simply join this RR operation with the best SPR operation in the SPRRS neighborhood.
Solving the TBRS problem
Similar to solving the SPRS problem, we can solve the TBRS problem given the solution to the TBRRS problem in the previous section.
Time complexity
We perform the same steps to solve the SPRS problem, and additionally by utilizing Lemmas 5 and 6 we compute and store the best rerooting for each subtree, which takes O(n^{3}) time. Overall, solving the TBRS problem still takes O(n^{3}) time.
Results and discussion
We examined the performance of our new triplet heuristics by comparing with two other supertree methods, MRP and modified MinCut (MMC; [20]), using published data sets from marsupials [21] and Cetartiodactyla [22]. MRP is the most widely applied supertree method [3]. However, MRP supertrees, like triplet supertrees, are intrinsically hard to compute. Therefore they are estimated using heuristics, which do not guarantee an optimal solution. In contrast, MMC supertrees can be computed exactly in polynomial time, and therefore, it has been suggested that MMC will be useful for building very large phylogenies [20]. We evaluated each of the supertree methods using the tripletsimilarity and the maximum agreement subtree (MAST) similarity [23] between the input trees and the supertrees. Furthermore, we measured the parsimony score of each computed supertree based on its binary matrix representation. For the marsupial data set, we also compared our results to published results using the max cut (MXC) supertree algorithm [24]. MXC is a modification of MMC that provides a heuristic approach based on the triplet supertree problem. There is currently no publicly available implementation of MXC (S. Snir, pers. comm.), and therefore, we were unable to apply it to the other data set.
Triplet supertrees were constructed using the programs TH(SPR) and TH(TBR) that implement hillclimbing heuristics based on our efficient local search algorithms for SPR and TBR branch swapping respectively. MRP supertrees were constructed using the parsimony hillclimbing heuristic implemented in PAUP* [9] with TBR branch swapping. We found very little difference in the results of MRP analyses when we collapsed zerolength branches and when we forced all MRP trees to be binary. We report the results from analyses that collapsed zerolength branches. All hillclimbing analyses were executed on 20 initial random addition sequence replicate trees and saving a single best tree per replicate. MMC supertrees were constructed using a program [20] supplied by Rod Page.
Our new triplet heuristics seek the supertrees with the maximum number of identical triplets to the collection of input trees, and indeed, we find they both outperform all other methods based on tripletsimilarity in all data sets (Table 1). Furthermore, all the triplet supertree analyses were completed within 15 minutes using a Kensington quadcore 2.66 GHz Linuxbased machine, demonstrating that our heuristics make the triplet supertree problem extremely tractable for largescale analyses. Both triplet heuristics and the MRP heuristic perform much better than the exact MMC algorithm, based on tripletsimilarity and MASTsimilarity. The MXC algorithm suffers from the lack of an available implementation; however, in our single comparison with the published results, this algorithm does not perform as well as either the MRP heuristic or our triplet heuristics.
Although the difference in the tripletsimilarity score between the triplet heuristics and the MRP heuristic is always less than 2% (Table 1), due to the extremely largenumber of triplets, even these apparently small differences represent large differences in tree topologies. For example, in the marsupial data set, the 0.7% difference in tripletsimilarity represents over 17,400 triplets. The comparison of the triplet heuristics and the MRP heuristic also demonstrates that optimizing the parsimony score of the matrix representation of input trees is not directly correlated with optimizing the tripletsimilarity of input trees and the supertree; supertrees with smaller (better) parsimony scores have lower (worse) tripletsimilarity (Table 1).
Our experiments also demonstrate that the supertree with the best tripletsimilarity is not necessarily best in terms of MASTsimilarity. In fact, MRP outperforms the triplet heuristics in terms of the MASTsimilarity to the input trees (Table 1). It is not intuitive that a parsimony analysis on a matrix representation of the input trees (MRP) is a valid or useful approach to infer the most accurate supertrees (but see [25]). The popularity of MRP is probably based more on the availability of programs that implement fast heuristics rather than a belief that parsimony on a matrix representation of input trees is the ideal supertree optimality criterion. However, MRP performs well in simple simulation experiments [26, 27] and in analyses of empirical data (e.g., [13]), and it clearly can be an effective supertree method. Since it is not obvious whether it is better to find supertrees that maximize accuracy in terms of tripletsimilarity, MASTsimilarity, or some other tree similarity measure like the RobinsonFoulds distance, we suggest that the triplet heuristics are an informative complement to the MRP method. Both approaches can provide supertrees that represent different, and equally valid, perspectives on accuracy.
Conclusion
Despite the inherent complexity of the triplet supertree problem, we have shown that it can be addressed effectively by using hillclimbing heuristics. We introduced efficient algorithms for standard local search problems that are solved by these heuristics. Our algorithms greatly improve on the best known (naïve) solutions for these search problems. This in turn makes hillclimbing heuristics for the triplet supertree problem applicable for largescale phylogenetic studies.
We demonstrate the performance of an implementation of our hillclimbing heuristics. In analyses of two empirical data sets, our triplet heuristics quickly found supertrees that contained more triplets in common with the input trees than supertrees found by MRP, MMC, or MXC. These results demonstrate not only that our heuristics for the triplet supertree problem make it a valuable alternative to standard supertree methods. They also demonstrate that developing new supertree heuristics that directly seek to optimize the accuracy of the supertree with respect to the input trees can enhance our ability to infer with accuracy large sections of the tree of life.
The algorithmic ideas developed in this work might set base for theoretical properties that identifies a much broader class of local search objectives, which can be solved more efficiently. This could lead to other powerful supertree heuristics. However, it remains an open problem if our solutions for the SPR and TBR based local search problems for the triplet supertree problem can be improved further.
References
 1.
Gordon AD: Consensus supertrees: The synthesis of rooted trees containing overlapping sets of labeled leaves. Journal of Classification. 1986, 3 (2): 335348.
 2.
Sanderson MJ, Purvis A, Henze C: Phylogenetic supertrees: assembling the trees of life. Trends in Ecology & Evolution. 1998, 13 (3): 105109.
 3.
BinindaEmonds ORP, Gittleman JL, Steel MA: The (super) tree of life: procedures, problems, and prospects. Annual Review of Ecology and Systematics. 2002, 33: 265289.
 4.
BinindaEmonds ORP: The evolution of supertrees. Trends in Ecology and Evolution. 2004, 19: 31522.
 5.
Davies JT, Barraclough TG, Chase MW, Soltis PS, Soltis DE, Savolainen V: Darwin's abominable mystery: Insights from a supertree of the angiosperms. PNAS. 2004, 101 (7): 19041909.
 6.
BinindaEmonds ORP, Cardillo M, Jones KE, MacPhee RDE, Beck RMD, Grenyer R, Price SA, Vos RA, Gittleman JL, Purvis A: The delayed rise of presentday mammals. Nature. 2007, 446 (7135): 507512.
 7.
Baum BR: Combining trees as a way of combining data sets for phylogenetic inference, and the desirability of combining gene trees. Taxon. 1992, 41: 310.
 8.
Ragan MA: Phylogenetic inference based on matrix representation of trees. Molecular Phylogenetics and Evolution. 1992, 1: 5358.
 9.
Swofford DL: PAUP*: Phylogenetic Analysis Using Parsimony (*and Other Methods). Version 4.0 beta. 2002, Sunderland, Massachusetts, USA: Sinauer Assoc
 10.
Goloboff PA: Techniques for analysis of large data sets. Techniques in Molecular Systematics and Evolution. Edited by: DeSalle R, Wheeler W, Giribet Ge. 2000, BirkhauserVerlag, Basel, 709.
 11.
Purvis A: A modification to Baum and Ragan's method for combining phylogenetic trees. Systematic Biology. 1995, 44: 2515.
 12.
Wilkinson M, Cotton JA, Creevey C: The shape of supertrees to come: tree shape related properties of fourteen supertree methods. Systematic Biology. 2005, 54: 41931.
 13.
Chen D, Eulenstein O, FernandezBaca D, Burleigh JG: Improved heuristics for minimumflip supertree construction. Evolutionary Bioinformatics. 2006, 2: 401410.
 14.
Bryant D: Building Trees, Hunting for Trees, and Comparing Trees – Theory and Methods in Phylogenetic Analysis. PhD thesis. 1997, University of Canterbury
 15.
Bansal M, Eulenstein O: The GeneDuplication Problem: NearLinear Time Algorithms for NNI Based Local Searches. Bioinformatics Research and Applications. 2008, 1425.
 16.
Page RDM, Holmes EC: Molecular evolution: a phylogenetic approach. 1998, Blackwell Science
 17.
Semple C, Steel M: Phylogenetics. 2003, Oxford University Press
 18.
Guigo R, Muchnik I, Smith TF: Reconstruction of ancient molecular phylogeny. Mol Phylogenet Evol. 1996, 6 (2): 189213.
 19.
Page RDM, Charleston M: From Gene to organismal phylogeny: reconciled trees and the gene tree/species tree problem. Mol Phylogenet Evol. 1997, 7: 231240.
 20.
Page RDM: Modified mincut supertrees. International Workshop, Algorithms in Bioinformatics (WABI). Edited by: Gusfield D, Guigó R. 2002, Lecture Notes in Computer Science, Springer Verlag, 2452: 300315.
 21.
Cardillo M, BinindaEmonds ORP, Boakes E, Purvis A: A specieslevel phylogenetic supertree of marsupials. Journal of Zoology. 2004, 264: 1131.
 22.
Price SA, BinindaEmonds ORP, Gittleman JL: A complete phylogeny of the whales, dolphins and eventoed hoofed mammals (Cetartiodactyla). Biological Reviews. 2005, 80 (3): 445473.
 23.
Burleigh JG, Eulenstein O, FernandezBaca D, Sanderson MJ: MRF supertrees. Phylogenetic supertrees: Combining Information to Reveal the Tree of Life. Edited by: BinindaEmonds ORP. 2004, Dordrecht: Kluwer Academic, 6585.
 24.
Snir S, Rao S: Using Max Cut to Enhance Rooted Trees Consistency. IEEE/ACM Trans Comput Biol Bioinformatics. 2006, 3 (4): 323333.
 25.
Bruen TC, Bryant D: Parsimony via Consensus. Systematic Biology. 2008, 57 (2): 251256.
 26.
BinindaEmonds ORP, Sanderson MJ: Assessment of the Accuracy of Matrix Representation with Parsimony Analysis Supertree Construction. Systematic Biology. 2001, 50 (4): 565579.
 27.
Eulenstein O, Chen D, Burleigh JG, FernandezBaca D, Sanderson M: Performance of flip supertrees with a heuristic algorithm. Systematic Biology. 2004, 53 (2): 299308.
Acknowledgements
This research is supported in part by National Science Foundation AToL grant EF0334832 and NESCent (NSF EF0423641).
This article has been published as part of BMC Bioinformatics Volume 10 Supplement 1, 2009: Proceedings of The Seventh Asia Pacific Bioinformatics Conference (APBC) 2009. The full contents of the supplement are available online at http://www.biomedcentral.com/14712105/10?issue=S1
Author information
Affiliations
Corresponding author
Additional information
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
HTL designed the triplet heuristics, implemented programs TH(SPR) and TH(TBR), and carried out the experiments. JGB led the analysis of the experimental results. OE inspired the triplet heuristics and supervised the project. All authors contributed to the writing of this manuscript, and have read and approved the final manuscript.
Rights and permissions
About this article
Cite this article
Lin, H.T., Burleigh, J.G. & Eulenstein, O. Triplet supertree heuristics for the tree of life. BMC Bioinformatics 10, S8 (2009). https://doi.org/10.1186/1471210510S1S8
Published:
Keywords
 Input Tree
 Root Edge
 Parsimony Score
 Supertree Method
 Local Search Problem