 Research
 Open Access
 Published:
A safe and complete algorithm for metagenomic assembly
Algorithms for Molecular Biology volume 13, Article number: 3 (2018)
Abstract
Background
Reconstructing the genome of a species from short fragments is one of the oldest bioinformatics problems. Metagenomic assembly is a variant of the problem asking to reconstruct the circular genomes of all bacterial species present in a sequencing sample. This problem can be naturally formulated as finding a collection of circular walks of a directed graph G that together cover all nodes, or edges, of G.
Approach
We address this problem with the “safe and complete” framework of Tomescu and Medvedev (Research in computational Molecular biology—20th annual conference, RECOMB 9649:152–163, 2016). An algorithm is called safe if it returns only those walks (also called safe) that appear as subwalk in all metagenomic assembly solutions for G. A safe algorithm is called complete if it returns all safe walks of G.
Results
We give graphtheoretic characterizations of the safe walks of G, and a safe and complete algorithm finding all safe walks of G. In the nodecovering case, our algorithm runs in time \(O(m^2 + n^3)\), and in the edgecovering case it runs in time \(O(m^2n)\); n and m denote the number of nodes and edges, respectively, of G. This algorithm constitutes the first theoretical tight upper bound on what can be safely assembled from metagenomic reads using this problem formulation.
Background
One of the oldest bioinformatics problems is to reconstruct the genome of an individual from short fragments sequenced from it, called reads (see [1,2,3] for some genome assembly surveys). Its most common mathematical formulations refer to an assembly (directed) graph built from the reads, such as a string graph [4, 5] or a de Bruijn graph [6, 7]. The nodes of such a graph are labeled with reads, or with substrings of the reads.^{Footnote 1} Standard assembly problem formulations require to find e.g., a nodecovering circular walk in this graph [8], an edgecovering circular walk [8,9,10,11],^{Footnote 2} a Hamiltonian cycle [12, 13] or an Eulerian cycle [7].
Real assembly graphs have however many possible solutions, due mainly to long repeated substrings of the genome. Thus, assembly programs used in practice, e.g., [5, 14,15,16,17,18], output only (partial) strings that are promised to occur in all solutions to the assembly problem. Following the terminology of [19], we will refer to such a partial output as a safe solution to an assembly problem; an algorithm outputting all safe solutions will be called complete. Even though practical assemblers incorporate various heuristics, they do have safe solutions at their core. Improving these can improve practical assembly results, and ultimately characterizing all safe solutions to an assembly problem formulation gives a tight upper bound on what can be reliably assembled from the reads.
We will assume here that the genome to be assembled is a node or edgecovering circular walk of the input graph, since Hamiltonian or Eulerian cycle formulations unrealistically assume that each position of the genome is sequenced exactly the same number of times. The quest for safe solutions for this assembly problem formulation has a long history. Its beginnings can be traced to [20], which assembled the paths whose internal nodes have indegree and outdegree equal to one. The method [7] assembled those paths whose internal nodes have outdegree equal to one, with no restriction on their indegree. Other strategies such as [9, 21, 22] are based on iteratively reducing the assembly graph, for example by contracting edges whose target has indegree equal to one. In [19], Tomescu and Medvedev found the first safe and complete algorithms for this problem, by giving a graphtheoretic characterization of all walks of a graph that are common to all of its node or edgecovering circular walks. The algorithm for finding them, though proven to work in polynomial time, launches an exhaustive visit of all walks starting at each edge, and extending each walk as long as it satisfies the graphtheoretic characterization.
The present paper is motivated by metagenomic sequencing [23, 24], namely the application of genomic sequencing to environment samples, such as soils, oceans, or parts of the human body. For example, metagenomic sequencing helped discover connections between bacteria in the human gut and bowel diseases [25, 26] or obesity [27]. A metagenomic sample contains reads from all the circular bacterial genomes present in it.
Because of the multiple genomes present in the sample, one can no longer formulate a solution for the metagenomic assembly problem as a single circular walk covering all the nodes or edges. A natural analog is to find a collection of circular walks of an assembly graph (i.e., the circular bacterial genomes), which together cover all the nodes, or edges, of the graph (i.e., they together explain all the reads). In general, we do not know how many bacterial species are in the sample, so we cannot place any bound on the number of circular walks. Hence, in our above formulation they can be any arbitrary number. See the next section for formal definitions, and Fig. 1 for a simple example.
It can be easily verified that the walks from [7, 9, 20,21,22]—which are safe for single circular covering walks—are also safe for this metagenomic problem formulation. However, even though many practical metagenomic assemblers exist, e.g., [28,29,30,31,32,33,34], no other safe solutions are known for this problem formulation.
In this paper we solve this problem, by giving a graphtheoretic characterization of all walks w of a graph G such that for any metagenomic assembly solution R of G, w is a subwalk of some circular walk in R. As opposed to the exhaustive search strategy from [19], in this paper we devise a new type of safe and complete algorithm for which we can tightly bound the running time. This works by outputting one solution to the metagenomic assembly problem, and then marking all its subwalks that satisfy our characterization. The algorithm for the nodecovering case can be implemented with a complexity of \(O(m^2 + n^3)\), and the one for the edgecovering case with a complexity of \(O(m^2n)\); n and m denote the number of nodes and edges, respectively, of the input graph. This is achieved by preprocessing the graph and the metagenomic assembly solution so that for each of its subwalks we can check in constant time if they satisfy our characterization.
We then show how to modify this algorithm to explicitly output all maximal safe walks (i.e., not contained in another safe walk), with a logarithmic slowdown, namely \(O(m^2 + n^3\log n)\) and \(O(m^2n\log n)\), respectively. This is based on constructing a suffixtree from the metagenomic assembly solution, and traversing it using suffix links.
Related work
This paper also falls into a broad line of research dealing with reallife problems that cannot model sufficiently well the real data. Other strategies for dealing with these in practice are to enumerate all solutions (as done e.g. in [35]), or to find the best k solutions (see e.g., [35, 36]).
Other bioinformatics studies that considered partial solutions common to all solutions are [37, 38], which studied basepairings common to all optimal alignments of two biological sequences under edit distance. In combinatorial optimization, safety has been studied under the name of persistency. For a given problem on undirected graphs, the persistent nodes or edges are those present in all solutions to the problem [39]. This question was first studied for the maximum matching problem of a bipartite graph [39], and later developed for more general assignment problems [40]. Later papers studied persistent nodes present in all maximum stable sets of a graph [41], or persistent edges present in all traveling salesman solutions on a particular class of graphs where the problem is polynomially solvable [42].
Persistency has been recently generalized from single edges to sets of edges by the notions of transversal and blocker [43]: a dtraversal is a set of edges intersecting any optimum solution in at least d elements, and a dblocker is a subset of edges whose removal deteriorates the value of the optimum solution by at least d. These notions have been studied for maximum matchings in arbitrary graphs [43], maximum stable sets [44], or for the maximum weight clique problem [45]. The problem closest to ours is the one of finding a minimumcardinality dtransversal of all s–t paths in a directed graph, shown to be polynomially solvable in [44].
Preliminaries and main definitions
In this paper by graph we always mean a directed graph. The number of nodes and edges in a graph G are denoted by n and m, respectively. We do not allow parallel edges, but allow selfloops and edges of opposite directions. For any node \(v \in V(G)\), we use \(N^(v)\) to denote its set of inneighbors, and \(N^+(v)\) to denote its set of outneighbors.
A walk in a graph is a sequence \(w = (v_0,e_0,v_1,e_1,\dots ,v_t,e_t,v_{t+1})\) where \(v_0,\dots ,v_{t+1}\) are nodes, and each \(e_i\) is an edge from \(v_i\) to \(v_{i+1}\) (\(t \ge 1\)). The length of w is its number of edges, namely \(t+1\). Walks of length at least one are called proper. Sometimes, we may omit explicitly writing the edges of w, and write only its nodes, i.e., \(w = (v_0,v_1,\dots ,v_t,v_{t+1})\). We will also say that an edge \((x,y) \in E(G)\) is a walk of length 1.
A path is a walk where all nodes are distinct. A walk whose first and last nodes coincide is called a circular walk. A path (walk) with first node u and last node v will be called a path (walk) from u to v, and will be denoted as uv path (walk). A cycle is a circular walk of length at least one (a selfloop) whose first and last nodes coincide, and all other nodes are distinct. If \(u = v\), then by v–u path we denote a cycle passing through v. A walk is called nodecovering or edgecovering if it passes through each node, or respectively edge, of the graph at least once.
Given a noncircular walk \(w = (v_0,v_1,\dots ,v_{t1})\) and a walk \(w' = (u_0,\dots ,u_{d1})\), we say that \(w'\) is a subwalk of w if there exists an index in w where an occurrence of \(w'\) starts. If \(w = (v_0,v_1,\dots ,v_{t1},v_t = v_0)\) is a circular walk, then we allow \(w'\) to “wrap around” w. More precisely, we say that \(w'\) is a subwalk of w if \(d \le t\) and there exists an index \(i \in \{0,\dots ,t1\}\) such that \(v_i = u_0\), \(v_{i+1 \bmod t} = u_1\), ..., \(v_{i+d1 \bmod t} = u_{d1}\).
The following reconstruction notion captures the notion of solution to the metagenomic assembly problem.
Definition 1
(Nodecovering metagenomic reconstruction) Given a graph G, a nodecovering metagenomic reconstruction of G is a collection R of circular walks in G, such that every node of G is covered by some walk in R.
The following definition captures the walks that appear in all nodecovering metagenomic reconstructions of a graph (see Fig. 1 for an example).
Definition 2
(Nodesafe walk) Let G be a graph with at least one nodecovering metagenomic reconstruction, and let w be a walk in G. We say that w is a nodesafe walk in G if for any nodecovering metagenomic reconstruction R of G, there exists a circular walk \(C \in R\) such that w is a subwalk of C.
We analogously define edgecovering metagenomic reconstructions and edgesafe walks of a graph G, by replacing node with edge throughout. Reconstructions consisting of exactly one circular nodecovering walk were considered in [19]. The following notion of nodeomnitig was shown in [19] to characterize the nodesafe walks of such reconstructions.
Definition 3
(Nodeomnitig, [19]) Let G be a graph and let \(w = (v_0,e_0,v_1,e_1,\dots ,v_t,e_t,v_{t+1})\) be a walk in G. We say that w is a nodeomnitig if both of the following conditions hold:

for all \(1 \le i \le j \le t\), there is no proper \(v_j\)–\(v_i\) path with first edge different from \(e_j\), and last edge different from \(e_{i1}\), and

for all \(0 \le j \le t\), the edge \(e_j\) is the only \(v_j\)–\(v_{j+1}\) path.
Theorem 1
[19] Let G be a strongly connected graph different from a cycle. A walk w in G is a subwalk of all nodecovering circular walks in G if and only if w is a nodeomnitig.
Observe that the circular walks in a nodecovering metagenomic reconstruction of a graph G stay inside its strongly connected components (because e.g., the graph of strongly connected components is acyclic). Likewise, a graph G admits at least one edgecovering metagenomic reconstruction if and only if G is made up of a disjoint union of strongly connected graphs. Thus, in the rest of the paper we will assume that the input graphs are strongly connected.
Characterizations of safe walks
In this section we give characterizations of node and edgesafe walks. The difference between our characterization below and Theorem 1 lies in the additional condition (b). Note that (b) refers to cycles, whereas the elements of a nodecovering metagenomic reconstruction are arbitrary circular walks; this is essential in our algorithm from the next section.
Theorem 2
Let G be a strongly connected graph. A walk \(w = (v_0,e_0,v_1,e_1,\dots ,v_t,e_t,v_{t+1})\) in G is a nodesafe walk in G if and only if the following conditions hold:

(a)
w is a nodeomnitig, and

(b)
there exists \(x \in V(G)\) such that w is a subwalk of all cycles passing through x.
Proof
\((\Rightarrow )\) Assume that w is safe. Suppose first that (a) does not hold, namely that w is not an omnitig. This implies that either (i) there exists a proper \(v_j\)\(v_i\) path p with \(1\le i \le j \le t\) with first edge different from \(e_j\), last edge different from \(e_{i1}\), or (ii) there exists j, \(0 \le j \le t\), and a \(v_j\)\(v_{j+1}\) path \(p'\) different from the edge \(e_j\).
Suppose (i) is true. For any nodecovering metagenomic reconstruction R of G, and any circular walk \(C \in R\) such that w is a subwalk of C, we replace C in R by the circular walk \(C'\), not containing w as subwalk, obtained as follows. Whenever C visits w until node \(v_j\), \(C'\) continues with the \(v_j\)–\(v_i\) path p, then it follows \((v_i,e_i,\dots ,e_{j1},v_j)\), and finally continues as C. Since p is proper, and its first edge is different from \(e_j\) and its last edge is different from \(e_{i1}\), the only way that w can appear in \(C'\) is as a subwalk of p. However, this implies that both \(v_j\) and \(v_i\) appear twice on p, contradicting the fact that p is a \(v_j\)–\(v_i\) path. Since each such circular walk \(C'\) covers the same nodes as C, the collection \(R'\) of circular walks obtained by performing all such replacements is also a nodecovering metagenomic reconstruction G. This contradicts the safety of w.
Suppose (ii) is true. As above, for any nodecovering metagenomic reconstruction R and any \(C \in R\) containing w as subwalk, we replace C with the circular walk \(C'\) obtained as follows. Whenever C traverses the edge \(e_j\), \(C'\) traverses instead \(p'\), and thus covers the same nodes as C, but does not contain w as subwalk. This also contradicts the safety of w.
Suppose now that (b) does not hold, namely, that for every \(x \in V(G)\), there exists a cycle \(c_x\) passing through x such that w is not a subwalk of \(c_x\). The set \(R = \{c_x \text{: } x \in V(G)\}\) is a nodecovering metagenomic reconstruction of G such that w is not a subwalk of any of its elements. This contradicts the safety of w.
\((\Leftarrow )\) Let R be a nodecovering metagenomic reconstruction of G, and let \(C \in R\) be a circular walk covering the vertex x. If C is a cycle, then (b) implies that w is a subwalk of C, from which the safety of w follows.
Otherwise, let G[C] be the subgraph of G induced by the edges of C. Clearly, C is a nodecovering circular walk of G[C], and thus G[C] is strongly connected. Moreover, we can argue that w is a nodeomnitig in G[C], as follows. By taking the shortest proper circular subwalk of C passing through x we obtain a cycle \(\widetilde{C}\) passing through x. From (b), we get that w is a subwalk of \(\widetilde{C}\). Since all edges of \(\widetilde{C}\) appear in G[C], then also all edges of w appear in G[C] and thus w is a walk in G[C]. The two conditions from the definition of nodeomnitigs are preserved under removing edges from G, thus w is a nodeomnitig also in G[C]. By applying Theorem 1 to G[C] we obtain that w is a subwalk of all nodecovering circular walks of G[C], and in particular, also of C. We have thus shown that for every nodecovering metagenomic reconstruction R of G, there exists \(C \in R\) such that w is a subwalk of C. Therefore, w is a nodesafe walk for G. \(\square\)
The following statement is a simple corollary of condition (b) from Theorem 2.
Corollary 3
Let G be a strongly connected graph, and let w be a safe walk in G. Then w is either a path or a cycle.
We now give the analogous characterization of edgesafe walks. We first recall the analogous definition of edgeomnitigs from [19]. This is the same as Definition 3, except that the second condition is missing.
Definition 4
(Edgeomnitig, [19]) Let G be a graph and let \(w = (v_0,e_0,v_1,e_1,\dots ,v_t,e_t,v_{t+1})\) be a walk in G. We say that w is an edgeomnitig if for all \(1 \le i \le j \le t\), there is no proper \(v_j\)–\(v_i\) path with first edge different from \(e_j\), and last edge different from \(e_{i1}\).
In [19], it was proven an equivalent of Theorem 1 in terms of edges, stating that edgeomnitigs characterize the walks of a strongly connected graph G that are subwalks of all edgecovering circular walks of G. Our characterization of the edgesafe walks considered in this paper is:
Theorem 4
Let G be a strongly connected graph. A walk \(w = (v_0,e_0,v_1,e_1,\dots ,v_t,e_t,v_{t+1})\) in G is an edgesafe walk in G if and only if the following conditions hold:

(a)
w is an edgeomnitig, and

(b)
there exists \(e \in E(G)\) such that w is a subwalk of all cycles passing through e.
Theorem 4 could be proved by carefully following the proof outline of Theorem 2. However, below we give a simpler proof, by reducing Theorem 4 to the nodecovering case in the graph S(G) obtained from G by subdividing every edge once.
Given a graph G, we let S(G) denote the graph obtained from G by subdividing each edge once. Namely, each edge (u, v) of G is replaced by two edges \((u,x_{uv})\), and \((x_{uv},v)\), where \(x_{uv}\) is a new node for every edge. Observe that the nodes \(x_{uv}\) have exactly one inneighbor, u, and exactly one outneighbor, v. We can analogously define this operation for a walk w in G, and then consider the walk S(w) in S(G).
Proof of Theorem 4
The proof follows the outline given in Fig. 2. We first argue that w is an edgesafe walk in G if and only if S(w) is a nodesafe walk in S(G). Indeed, observe that the edgecovering metagenomic reconstructions of G are in bijection with the nodecovering metagenomic reconstructions of S(G), the bijection being \(R \mapsto \{S(C) \text{: } C \in R\}\). Moreover, w is a subwalk of a walk C in G if and only if S(w) is a subwalk of S(C) in S(G). Therefore, w is an edgesafe walk in G if and only if S(w) is a nodesafe walk in S(G).
It remains to show that w satisfies conditions (a) and (b) of Theorem 4 for G if and only if S(w) satisfies conditions (a) and (b) of Theorem 2 for S(G).
Condition (a): It immediately follows from the definition that if S(w) is a nodeomnitig in S(G) then w is an edgeomnitig in G. Assume now that w is an edgeomnitig in G. By the construction of S(G) and S(w), between any two consecutive nodes of S(w) there can be only one path in S(G) (namely, the edge connecting the two nodes). Therefore, S(w) is a nodeomnitig in S(G).
Condition (b): Suppose that there exists an edge \(e = (u,v) \in E(G)\) such that all cycles in G passing through e contain w as subwalk. Then by construction all cycles in S(G) passing through \(x_{uv} \in V(S(G))\) also contain S(w) as subwalk. Conversely, suppose that there exists a node \(x \in V(S(G))\) such that all cycles in S(G) passing through x contain S(w) as subwalk. If x is a node of the type \(x_{uv}\) for some edge (u, v) of G, then it also holds that all cycles in G passing through \((u,v) \in E(G)\) contain w as subwalk. Otherwise, if \(x \in V(G)\), then let (x, y) be an arbitrary edge of G outgoing from x; this exists because G is strongly connected. We claim that all cycles in G passing through \((x,y) \in E(G)\) contain w as subwalk. Indeed, let \(z_{xy}\) be the node of S(G) corresponding to the edge (x, y). The set of cycles of S(G) passing through \(z_{xy}\) is a subset of the set of cycles of S(G) passing through x. Therefore, all cycles of S(G) passing through \(z_{xy}\) contain S(w) as subwalk. We have now reduced this case to the previous one, when x is a node of the type \(x_{uv}\) for some edge (u, v) of G, and the claim follows. \(\square\)
The algorithm for finding all nodesafe walks
In this section we give an algorithm for finding all nodesafe walks of a strongly connected graph. In the next section we show how to implement this algorithm to run in \(O(m^2+n^3)\) time. Our results for edgesafe walks are analogous, and will be given in the last section.
We begin with an easy lemma stating a simple condition when a maximum overlap of two nodeomnitigs is a nodeomnitig.
Lemma 5
Let G be a graph, and let \(w = (v_0,e_0,v_1,\dots ,v_t,e_t,v_{t+1})\) be a walk of length at least two in G. We have that w is a nodeomnitig if and only if \(w_1 = (v_0,e_0,v_1,\dots ,v_t)\) and \(w_2 = (v_1,e_1,v_2,\dots ,v_t,e_t,v_{t+1})\) are nodeomnitigs and there is no \(v_t\)–\(v_1\) path with first edge different than \(e_t\) and last edge different than \(e_0\).
Proof
The forward implication is trivial, as by definition subwalks of nodeomnitigs are nodeomnitigs. For the backward implication, since both \(w_1\) and \(w_2\) are nodeomnitigs, then for all \(0 \le j \le t\), the edge \(e_j\) is the only \(v_j\)–\(v_{j+1}\) path. Since \(w_1\) is a nodeomnitig, then for all \(1 \le i \le j \le t1\), there is no proper \(v_j\)\(v_i\) path with first edge different from \(e_j\), and last edge different from \(e_{i1}\). If there is no \(v_t\)\(v_1\) path with first edge different than \(e_t\) and last edge different than \(e_0\), we obtain that w is a nodeomnitig. \(\square\)
The following definition captures condition (b) from Theorem 2. Note that the walk w can also be a single node.
Definition 5
(Certificate) Let G be a graph and let w be a walk in G. A node \(x \in V(G)\) such that w is a subwalk of all cycles passing through x is called a certificate of w. The set of all certificates of w will be denoted \(\mathsf {Cert}(w)\).
By Theorem 2, nodesafe walks are those nodeomnitigs with at least one certificate. In the following lemma we relate the certificates of a nodeomnitig with the certificates of its nodes. Later, in Lemma 8, we will show that the certificates of single nodes can be computed efficiently.
Lemma 6
Let G be a graph and let \(w = (v_0,e_0,v_1,\dots ,v_t,e_t,v_{t+1})\) be a proper nodeomnitig in G. Then \(\mathsf {Cert}(w) = \mathsf {Cert}(v_0) \cap \mathsf {Cert}(v_1) \cap \cdots \cap \mathsf {Cert}(v_{t+1})\).
Proof
We prove the claim by doubleinclusion. The inclusion \(\mathsf {Cert}(w) \subseteq \mathsf {Cert}(v_0) \cap \mathsf {Cert}(v_1) \cap \cdots \cap \mathsf {Cert}(v_{t+1})\) is trivial, since all cycles passing through a node \(x \in \mathsf {Cert}(w)\) also contain each of \(v_0,\dots ,v_{t+1}\).
We now prove the reverse inclusion by induction on the length of w. We first check the base case when w has length one. Assume for a contradiction that there is a cycle C passing through \(x \in \mathsf {Cert}(v_0) \cap \mathsf {Cert}(v_1)\) and not having \(w = (v_0,e_0,v_1)\) as subpath. Then, after visiting x, (i) C first traverses \(v_0\) and then reaches \(v_1\) with a path different than \(e_0\), or (ii) C first traverses \(v_1\) and then \(v_0\). The case (i) is immediately excluded, since w is a nodeomnitig and \(e_0\) is the only \(v_0\)–\(v_1\). If (ii) holds, then there is a x\(v_1\) path \(P_1\) and a \(v_0\)x path \(P_0\). However, the concatenation of \(P_0\) with \(P_1\) is a \(v_0\)\(v_1\) path different than the edge \(e_0\), which again contradicts the fact that w is a nodeomnitig.
We now use the inductive hypothesis to show that if \(x \in \mathsf {Cert}(v_0) \cap \mathsf {Cert}(v_1) \cap \cdots \cap \mathsf {Cert}(v_{t+1})\), then \(x \in \mathsf {Cert}(w)\). We partition w into the two walks \(w_0 = (v_0,e_0,v_1,\dots ,v_{t})\) and \(w_t = (v_t,e_t,v_{t+1})\). By induction, since \(x \in \mathsf {Cert}(v_0) \cap \mathsf {Cert}(v_1) \cap \cdots \cap \mathsf {Cert}(v_t)\) we have \(x \in \mathsf {Cert}(w_0)\). Analogously, since \(x \in \mathsf {Cert}(v_t) \cap \mathsf {Cert}(v_{t+1})\), we have \(x \in \mathsf {Cert}(w_t)\). Since \(v_t\) is a node in both \(w_0\) and \(w_t\), then any cycle passing through x, once it passes through \(w_0\) it must continue passing through \(w_t\). Therefore, any cycle passing through x passes also through w, and hence \(x \in \mathsf {Cert}(w)\). \(\square\)
Given a circular walk \(C = (v_0,e_0,v_1,\dots ,v_{d1},e_{d1},v_d = v_0)\), \(i \in \{0,\dots ,d1\}\) and \(k \in \{0,\dots ,d\}\), we denote by C(i, k) the subwalk of C starting at \(v_i\) and of length k, that is, \(C(i,k) = (v_i,e_i,v_{i+1 \bmod d},\dots ,v_{(i+k) \bmod d})\).
Algorithm 1 finds all nodesafe walks of a strongly connected graph G (possibly with duplicates), but does not return each nodesafe walk explicitly. Instead, it returns a nodecovering circular walk C of G and the set of pairs (i, k) such that C(i, k) is a nodesafe walk.
The algorithm works by scanning C and checking whether each subwalk of C starting at index i and of length k is a nodeomnitig and has at least one certificate. If so, then it stores the index i in a set \(S_k\), for every k. The algorithm first deals with the case \(k=1\): it first checks whether C(i, 1) is a nodeomnitig (Line 7) and whether it has at least one certificate (Line 8). The case \(k > 1\) is analogous. It first checks whether \(C(i,k1)\) and \(C(i+1 \bmod d,k1)\) are omnitigs (by checking the memberships \(i \in S_{k1}\) and \(i+1 \bmod d \in S_{k1}\)) and that there is no path as in the definition of nodeomnitig (Line 11). Then it checks whether C(i, k) has at least one certificate (Line 12).
In the next section we show how to preprocess G and C to perform these verifications in constant time. This algorithm can be modified to output nodesafe walks also without duplicates. For clarity, we explain this idea in the proof of Theorem 13, where we also show how to output only maximal nodesafe walks, i.e., those that are not subwalks of any other nodesafe walk.
Theorem 7
Given a strongly connected graph G, Algorithm 1 correctly computes all the nodesafe walks of G, possibly with duplicates.
Proof
We will first prove by induction on k that the set \(S_k\) contains all those indices i for which C(i, k) is a nodesafe walk of length k. In the base case \(k = 1\), we explicitly check if each C(i, 1) is a nodeomnitig (Line 7). We also check if C(i, 1) has at least one certificate, by checking (due to Lemma 6) whether \(\mathsf {Cert}(v_i) \cap \mathsf {Cert}(v_{i+1 \bmod 1}) \ne \emptyset\) (Line 8). Thus, for each i we checked whether C(i, 1) is a nodesafe walk (due to Theorem 2), and the claim follows for \(S_1\). We assume now that the claim is true for \(S_{k1}\). For each i, by Lemma 5, C(i, k) is a nodeomnitig if and only if \(C(i,k1)\) and \(C(i+1 \bmod d,k1)\) are nodeomnitigs, and there is no \(v_{i+k1 \bmod d}\)\(v_{i+1 \bmod d}\) path with first edge different than \(e_{i+k1 \bmod d}\) and last edge different than \(e_i\). This is verified in Line 11. In Line 12 we check whether \(\mathsf {Cert}(C(i,k)) \ne \emptyset\) by checking whether \(\mathsf {Cert}(v_i) \cap \dots \cap \mathsf {Cert}(v_{i+k \bmod d}) \ne \emptyset\) (due to Lemma 6). Thus the claim is true for all \(S_k\).
By Corollary 3, all nodesafe walks of G are paths or cycles, thus of length at most n. By the definition of nodesafe, they are also subwalks of C. Thus for each nodesafe walk w of G of length \(k \le n\), there exists \(i \in \{0,\dots ,d1\}\) such that \(w = C(i,k)\) and \(i \in S_k\). \(\square\)
An \(O(m^2 + n^3)\) implementation for nodesafe walks
In this section we describe the implementation of Algorithm 1. We first show how to compute the certificates of all nodes.
Lemma 8
Let G be a strongly connected graph with n nodes and m edges. We can compute the sets \(\mathsf {Cert}(x)\) for all, in time \(x \in V(G)\)O(mn).
Proof
We start by initializing \(\mathsf {Cert}(x) = \{x\}\) for every node x (recall that G is strongly connected). We then construct the graph \(G'\) by subdividing every node of G once. That is, we replace every node x of G with two nodes \(x_{in}\) and \(x_{out}\), and add the edge \((x_{in},x_{out})\) to \(G'\). Moreover, for every edge (y, z) of G, we add to \(G'\) the edge \((y_{out},z_{in})\). Observe that also \(G'\) is strongly connected.
For every \(x \in V(G)\), we compute \(\mathsf {Cert}(x)\) as follows. We consider the graph \(G'_x\) obtained from \(G'\) by removing the edge \((x_\text{{in}},x_\text{{out}})\). We compute the strongly connected components of \(G'_x\), in time O(m). We then iterate through all \(y \in V(G) \setminus \{x\}\) and check in constant time whether \(y_\text{{in}}\) and \(y_\text{{out}}\) still belong to the same strongly connected component of \(G'_x\). If not, then x belongs to all cycles of G passing through y, and thus we add y to \(\mathsf {Cert}(x)\). This takes in total O(mn) time. \(\square\)
The following lemma shows how to check in constant time the first condition in the definition of nodeomnitig.
Lemma 9
Let G be a graph with m edges. We can preprocess G in time \(O(m^2)\) and space \(O(m^2)\) such that for every two distinct edges, \((x_1,y_1),(x_2,y_2) \in E(G)\) we can answer in O(1) time if there is a \(x_1\)–\(y_2\) path in G with first edge different than \((x_1,y_1)\) and last edge different than \((x_2,y_2)\).
Proof
We show how to precompute a table \(a(\cdot ,\cdot )\) of size \(O(m^2)\) that for any two distinct edges \((x_1,y_1),(x_2,y_2) \in E(G)\) stores the answer to the query. See Fig. 3 for an illustration.
We iterate through all edges \((x_1,y_1) \in E(G)\), and consider the graph \(G_{(x_1,y_1)}\) obtained from G by removing \((x_1,y_1)\). We launch a graph visit in \(G_{(x_1,y_1)}\) from \(x_1\) to compute to which nodes there is a path from \(x_1\). By construction, any such path starts with an edge different than \((x_1,y_1)\). We then consider each node \(z \in V(G)\). We first iterate once through the inneighbors of z to compute how many of its inneighbors are reachable from \(x_1\) in \(G_{(x_1,y_1)}\); say this number is \(d_z\). We then iterate a second time through the inneighbors of z, and for each inneighbor w, we let \(r_w\) be equal to 1 if w is reachable from \(x_1\) in \(G_{(x_1,y_1)}\), and 0 otherwise. We have that there is a \(x_1\)z path in G with first edge different than \((x_1,y_1)\) and last edge different than (w, z) if and only if \(d_z  r_w > 0\). Thus we set
The complexity of this algorithm is \(O(m^2)\), because for every edge \((x_1,y_1)\), we compute the set of nodes reachable from \(x_1\) in time O(m), and then we process each edge of \(G_{(x_1,y_1)}\) exactly two times. \(\square\)
Using e.g., the result of [46], we can also verify the second condition in the definition of nodeomnitig in constant time.
Lemma 10
Let G be a graph with m edges, we can preprocess G in time O(m) such that for every edge \((x,y) \in E(G)\) we can answer in O(1) time whether (x, y) is the only x–y path .
Proof
A strong bridge is an edge whose removal increases the number of strongly connected components of a graph (see e.g., [46]). It is easy to see that an edge \((x,y) \in E(G)\) is the only x–y path if and only if (x, y) is a strong bridge. In [46] it was shown that all strong bridges can be computed in linear time in the size of the graph, from which our claim follows. \(\square\)
The following lemma shows how to check in constant time condition (b) from Theorem 2. The idea is to precompute, for every index i in C, the smallest (i.e., leftmost) index \(i  n \le \ell (i) \le i\) such that \(\mathsf {Cert}(v_{\ell (i)}) \cap \dots \cap \mathsf {Cert}(v_{i}) \ne \emptyset\). C(i, k) has a nonempty set of certificates if and only if \(\ell (i)\) is at distance at least k to i, that is, \(k \le i \ell (i)\).
Lemma 11
Let G be a graph with n nodes and m edges, and let \(C = (v_0,e_0,v_1,\dots ,v_{d1},e_{d1},v_d = v_0)\) be a circular walk in G, with \(n \le d \le n^2\). We can preprocess G and C in time , such that for every \(O(n^3)\) \(i \in \{0,\dots ,d1\}\) and, we can answer in \(k \in \{0,\dots ,n\}\) O(1) time if \(\mathsf {Cert}(v_i) \cap \dots \cap \mathsf {Cert}(v_{i+k \bmod d}) \ne \emptyset\).
Proof
To simplify the notation, given an integer i, by \(v_i\) we always mean \(v_{i \bmod d}\). By Lemma 8, we can compute \(\mathsf {Cert}(x)\), for every \(x \in V(G)\), in \(O(mn) \in O(n^3)\) time. In addition to computing the index \(\ell (i)\), we also compute the intersection \(L_i = \mathsf {Cert}(v_{\ell (i)}) \cap \dots \cap \mathsf {Cert}(v_{i})\). Each such intersection set is stored as an array of length n telling in how many of \(\mathsf {Cert}(v_{\ell (i)}), \dots ,\mathsf {Cert}(v_{i})\) each \(x \in V(G)\) is contained; \(L_i\) is nonempty if and only if there is an entry in this array with a value equaling the number of sets \(\mathsf {Cert}(v_{\ell (i)}), \dots , \mathsf {Cert}(v_{i})\).
We begin by computing \(\ell (i)\) and \(L_i\) for \(i = 0\) in a straightforward manner, by trying \(\ell (i) = t = i  1, i2,\ldots\) as long as the resulting intersection is nonempty. Namely, we initialize \(L_i = \mathsf {Cert}(v_i)\), and update it as \(L_i := L_i \cap \mathsf {Cert}(v_{t})\). We keep decreasing t as long as \(L_i\) is nonempty. If t reaches 0, then all sets \(\mathsf {Cert}(x)\) have a common element, and the answer is “yes” for any query. Computing each intersection takes time O(n), and there are O(d) intersections to compute, giving a total of \(O(dn) \in O(n^3)\) time.
For \(i > 0\), we compute \(\ell (i)\) as follows. We first compute \(L_{i1} \cap \mathsf {Cert}(v_i)\). If this is nonempty, then \(L_i := L_{i1} \cap \mathsf {Cert}(v_i)\) and \(\ell (i) := \ell (i1)\). By the way we store intersection sets, this can be done in O(n) time.
Otherwise, we keep increasing \(\ell (i)\) by one from \(t = \ell (i1)\) until the corresponding intersection \(\mathsf {Cert}(v_{t}) \cap \dots \cap \mathsf {Cert}(v_{i})\) is nonempty. We then set \(L_i\) to this intersection and \(\ell (i) = t\). By the way we store the intersections, it follows that we can compute the new intersection in time O(n), by scanning the current intersection and removing the elements of \(\mathsf {Cert}(v_{t})\) from it, by decreasing by one the counters of its elements. Overall, such new intersections are computed at most d times, because for each i we start this scan from index \(\ell (i1)\) onwards, and always \(\ell (i1) \le \ell (i) \le i\) holds. This gives a total complexity of \(O(nd) \in O(n^3)\). \(\square\)
We are now ready to combine these lemmas into the main theorem of this section.
Theorem 12
Algorithm 1 can be implemented to run in time \(O(m^2 + n^3)\) for any strongly connected graph with n nodes and m edges.
Proof
Any strongly connected graph admits a nodecovering circular walk \(C = (v_0,e_0,v_1,\dots ,v_{d1},e_{d1},v_d = v_0)\) of length \(d \in \{n,\dots ,n^2\}\), that can be constructed in time \(O(nm) \in O(n^3)\). For example, one can label the nodes of G as \(v_1,\dots ,v_n\), start at \(v_1\), then follow an arbitrary path until \(v_2\) (which exists since G is strongly connected), and then continue from \(v_2\) in the same manner. This is the same argument given in [19].
By Lemma 8, we can compute in time \(O(mn) \in O(n^3)\) the sets \(\mathsf {Cert}(x)\) for all \(x \in V(G)\). We preprocess G and C as indicated in Lemmas 9, 10, and 11, in time \(O(m^2 + n^3)\). For every length \(k \in \{1,\dots ,n\}\), and every index \(i \in \{0,\dots ,d1\}\), this allows us to perform all checks in constant time. Checking membership to \(S_{k1}\) can also be done in constant time, by storing each set \(S_{k}\) as a bitvector of length d. \(\square\)
In the next section we discuss how to optimize Algorithm 1 to start with a nodecovering metagenomic reconstruction of minimum total length. However, there are graphs in which any nodecovering metagenomic reconstruction has length \(\Omega (n^2)\), see Fig. 4.
Additional results
Maximal nodesafe walks without duplicates
In a practical genome assembly setting we want to reconstruct fragments of the genomes as long as possible. As such, we are interested only in maximal nodesafe walks, that is, in nodesafe walks that are not subwalks of any other nodesafe walk. A trivial way to obtain these is to take the output of Algorithm 1, convert it into the set of all nodesafe walks of G, and run a suffixtreebased algorithm for removing the nonmaximal ones, in time linear in their total length. However, given a nodecovering circular walk C of length \(d \le n^2\), the total length of the nodesafe walks is at most \(\sum _{k = 0}^{n}kd \in O(n^4)\).
In the next theorem we show how to reduce this time complexity to \(O(m^2+n^3\log n)\). The main observation is that a nodesafe walk of length k is maximal if it is not extended into a nodesafe walk of length \(k+1\). We avoiding outputting duplicate maximal walks by traversing a suffixtree built from C to check for previous occurrences of each walk of length k.
Theorem 13
Given a strongly connected graph G with n nodes and m edges, Algorithm 1 can be modified to output the maximal nodesafe walks of G explicitly and without duplicates, with a running time of \(O(m^2 + n^3\log n)\).
Proof
Let \(C = (v_0,\dots ,v_{d} = v_0)\) be a nodecovering circular walk C of G, of length \(n \le d \le n^2\). At any position in C there can start the occurrence of at most one maximal nodesafe walk. By Corollary 3, the length of each nodesafe walk is at most n, thus the sum of the lengths of all maximal nodesafe walks of G is \(O(n^3)\). This implies that if we find the occurrences in C of all maximal nodesafe walks without duplicates, then we can output all of them explicitly within the stated time bound.
A nodesafe walk w of length k is maximal if no occurrence C(i, k) of w in C was extended left or right at step \(k+1\). We can keep track of all previous occurrences of w in C, as follows. Initially, we build the suffix tree T of the (linear) string \(C' = v_0v_1\ldots v_{d}v_1\ldots v_{n2}\#\) over the alphabet \(\Sigma = V(G) \cup \{\#\}\), where \(\#\) is a new symbol. This takes time linear in the size of \(C'\) and in the alphabet size \(\Sigma  = n\), thus \(O(n^2)\) [47]. As we scan C for a length \(k+1 \in \{1,\dots ,n\}\), we maintain, as we discuss below, a pointer in T to the node \(u_i\) such that the label of the path from the root to \(u_i\) spells C(i, k). In \(u_i\) we store the information of whether any occurrence the walk \(w = C(i,k)\) was extended at step \(k+1\).
As we advance from i to \(i+1\), we follow a socalled suffixlink in T to move to the node \(u^*\) such that the label from the root of T to \(u^*\) spells \(C(i+1,k1)\) (i.e., C(i, k) with its first character removed). For a detailed discussion on the properties of the suffix tree, see e.g., [48]. We then follow the normal tree edge exiting from \(u^*\) labeled \(v_{i+1 \bmod d}\). We thus advance to the node \(u_{i+1}\) of T such that the path from the root to \(u_{i+1}\) spells \(C(i+1,k)\). See Fig. 5 for an illustration. After traversing once C at step \(k+1\) and detecting which nodesafe walks of length k are maximal, we traverse C again to output these nodesafe walk.
After building the suffix tree using [47], the children of each node are organized in lexicographic order. Descending in the tree takes at most \(O(\log (\Sigma )) = O(\log n)\) time per step for binary searching the first character of each edge. Following suffix links can be be amortized to the descending operations [48]. Thus, the above additional phase takes time \(O(n^3\log n)\). The precomputations needed in the proof of Theorem 12 take time \(O(m^2 + n^3)\), from which the claimed time complexity bound follows. \(\square\)
The algorithm for finding edgesafe walks
In this section we adapt Algorithm 1 and its implementation to find edgesafe walks, as characterized by Theorem 4. This will result in an algorithm running in time \(O(m^2n)\). The proof of the following theorem is entirely analogous to the nodesafe case.
Theorem 14
Let G be a strongly connected graph with n nodes and m edges. In time we can output an edgecovering circular walk \(O(m^2n)\) C, and the set of all pairs (i, k) such that C(i, k) is an edgesafe walk of G.
Proof
The proof is analogous to the nodesafe case, and thus we briefly sketch the differences. In the edgecovering case, the set of certificates of a walk w consists of the edges e such that all cycles passing through e contain w as subwalk. Analogously to Lemma 6, we have that the set of certificates of a walk w equals the intersection of the sets of certificates of its individual edges. The algorithm for the edgesafe case is that same as Algorithm 1, with the difference that we now start with an edgecovering circular walk C and we do not check anymore that each C(i, 1) is the only \(v_i\)–\(v_{i+1}\) path.
By the same argument given in the proof of Theorem 12, such a circular walk C has length at most mn and can be found in time O(mn). The certificates of all edges can be similarly computed in time \(O(m^2)\) (now there is no need to subdivide nodes into single edges). Lemma 9 can be applied verbatim without modifications. The analog of Lemma 11 now starts with an edgecovering circular walk C of length O(mn). The only difference in its proof is that the sets of certificates now have size at most m, thus their intersection takes time O(m). This implies that we can precompute G and C in time \(O(m^2n)\).
After this preprocessing phase, the algorithm itself works in time \(O(mn^2)\), since the edgecovering circular walk C has length O(mn). \(\square\)
With a proof identical to the one of Theorem 13, we also obtain the following result.
Theorem 15
Given a strongly connected graph G with n nodes and m edges, we can output the maximal edgesafe walks of G explicitly and without duplicates, in time of \(O(m^2n\log n)\).
Optimizations to the algorithms
A trivial way to optimize Algorithm 1 is to start with a nodecovering circular walk of minimum length. However, this is an NPhard problem, since G has a nodecovering circular walk of length n if and only if G is Hamiltonian. Observe though that instead of a single nodecovering circular walk, we can start with a nodecovering metagenomic reconstruction possibly consisting of multiple circular walks, and apply Algorithm 1 to each walk in the reconstruction. This is correct by definition, since nodesafe walks are subwalks of some walk in any nodecovering metagenomic reconstruction.
Finding a nodecovering metagenomic reconstruction whose circular walks have minimum total length can be solved with a minimumcost circulation problem (see e.g., [49, 50] for basic results on minimumcost circulations). From G, we construct the graph \(G'\) by subdividing every node of G once (recall the construction from Lemma 8). We set demand 1 and cost 0 on each edge \((x_\text{{in}},x_\text{{out}})\), with \(x \in V(G)\). On all edges corresponding to original edges of G we set demand 0 and cost 1. A circulation f in \(G'\) satisfying the demands can be decomposed into cycles, which form a nodecovering metagenomic reconstruction in G. The total length of these cycles in G equals the cost of f. Since \(G'\) has no capacities, a minimumcost circulation in \(G'\) can be found in time \(O(n\log U(m + n\log n))\), where U is the maximum value of a demand, using the algorithm of Gabow and Tarjan [51]. All demands in \(G'\) are 1, thus this bound becomes \(O(nm + n^2\log n)\).
In the algorithm for finding all edgecovering circular walks, we need to find an edgereconstruction whose circular walks have minimum total length. This can be solved as above, without subdividing the nodes of G. We add to every edge the demand 1 and cost 1 and then compute a minimumcost circulation. The decomposition of the optimal circulation into cycles forms an edgereconstruction of G.
Conclusions and future work
We consider [19] and the present work as starting points for characterizing all safe solutions for natural assembly problem formulations, and thus obtaining safe and complete algorithms.
As future work, we plan to study formulations where the assembly solution is made up of noncircular covering walks, or where the assembly solutions consist of a given number of covering walks (e.g., a given number of chromosomes). In terms of real graph instances, the worstcase complexity of our algorithm may be prohibitive, and thus improving it is an important problem.
We also leave for future work an idealized experimental study similar to what was done for the single genome case in [19]. This compared the lengths and biological content of some classes of safe solutions known in the literature, on de Bruijn graphs constructed from errorfree, singlestranded simulated reads.
The ultimate goal of a “safe and complete” approach is to be adapted to the peculiarities of real data, such as sequencing errors, insufficient sequencing coverage, reverse complements. However, our belief is that we first need a clean and solid theoretical foundation, which can later be extended, or weakened, to account for such features.
Notes
 1.
 2.
Node and edgecovering walks usually refer to node and edgecentric de Bruijn graphs, respectively. In the nodecentric de Buijn graph, all kmers in the reads are nodes of the graph, and edges are added between all kmers that have a suffixprefix overlap of length \(k1\). In the edgecentric de Bruijn graph, it is further required that the \(k+1\)mer obtained by overlapping the two kmers of an edge also appears in the reads. Thus for edgecentric de Bruijn graphs it reasonable to require that the walk covers all edges, because all edges also appear in the reads; this may not be the case for nodecentric de Bruijn graphs.
References
 1.
Miller JR, Koren S, Sutton G. Assembly algorithms for nextgeneration sequencing data. Genomics. 2010;95(6):315–27.
 2.
Nagarajan N, Pop M. Sequence assembly demystified. Nat Rev Genet. 2013;14(3):157–67.
 3.
Simpson JT, Pop M. The theory and practice of genome sequence assembly. Annu Rev Genom Hum Genet. 2015;16:153–62. https://doi.org/10.1146/annurevgenom090314050032.
 4.
Myers EW. The fragment assembly string graph. Bioinformatics. 2005;21(suppl–2):79–85.
 5.
Simpson JT, Durbin R. Efficient de novo assembly of large genomes using compressed data structures. Genome Res. 2011;22(3):549–56.
 6.
Idury RM, Waterman MS. A new algorithm for DNA sequence assembly. J Comput Biol. 1995;2(2):291–306.
 7.
Pevzner PA, Tang H, Waterman MS. An Eulerian path approach to DNA fragment assembly. Proc Nat Acad Sci. 2001;98:9748–53.
 8.
Nagarajan N, Pop M. Parametric complexity of sequence assembly: theory and applications to next generation sequencing. J Comput Biol. 2009;16(7):897–908.
 9.
Medvedev P, Brudno M. Maximum likelihood genome assembly. J Comput Biol. 2009;16(8):1101–16.
 10.
Medvedev P, Georgiou K, Myers G, Brudno M. Computability of models for sequence assembly. WABI. 2007;4645:289–301.
 11.
Kapun E, Tsarev F. De Bruijn superwalk with multiplicities problem is NPhard. BMC Bioinform. 2013;14(Suppl 5):7.
 12.
Lysov IP, Florent’ev VL, Khorlin AA, Khrapko KR, Shik VV. Determination of the nucleotide sequence of DNA using hybridization with oligonucleotides. A new method. Doklady Akademii nauk SSSR. 1988;303(6):1508–11.
 13.
Narzisi G, Mishra B, Schatz MC. On algorithmic complexity of biomolecular sequence assembly problem. Algorithms for computational biology. 2014. Springer, Cham, p. 183–95.
 14.
Simpson JT, Wong K, Jackman SD, Schein JE, Jones SJ, Birol İ. ABySS: a parallel assembler for short read sequence data. Genome Res. 2009;19(6):1117–23.
 15.
Butler J, Maccallum I, Kleber M, Shlyakhter IA, Belmonte MK, Lander ES, Nusbaum C, Jaffe DB. Allpaths: de novo assembly of wholegenome shotgun microreads. Genome Res. 2008;18(5):810–20.
 16.
Li R, Zhu H, Ruan J, Qian W, Fang X, Shi Z, Li Y, Li S, Shan G, Kristiansen K. De novo assembly of human genomes with massively parallel short read sequencing. Genome Res. 2010;20(2):265.
 17.
Iqbal Z, Caccamo M, Turner I, Flicek P, McVean G. De novo assembly and genotyping of variants using colored de Bruijn graphs. Nat Genet. 2012;44(2):226–32.
 18.
Zerbino DR, Birney E. Velvet: algorithms for de novo short read assembly using de Bruijn graphs. Genome Res. 2008;18(5):821–9.
 19.
Tomescu AI, Medvedev P. Safe and complete contig assembly via omnitigs. In: Singh M. (ed.). Research in computational molecular biology20th annual conference, RECOMB 2016, Santa Monica, CA, USA, April 17–21, 2016. In: Proceedings lecture notes in computer science. 2016, vol 9649, p. 152– 63. Springer, cham. https://doi.org/10.1007/9783319319575.
 20.
Kececioglu JD, Myers EW. Combinatiorial algorithms for DNA sequence assembly. Algorithmica. 1995;13(1/2):7–51.
 21.
Jackson BG. Parallel methods for short read assembly. Ph.D. Thesis, Iowa State University. 2009.
 22.
Kingsford C, Schatz MC, Pop M. Assembly complexity of prokaryotic genomes using short reads. BMC Bioinform. 2010;11(1):21.
 23.
Venter JC, Remington K, Heidelberg JF, Halpern AL, Rusch D, Eisen JA, Wu D, Paulsen I, Nelson KE, Nelson W, Fouts DE, Levy S, Knap AH, Lomas MW, Nealson K, White O, Peterson J, Hoffman J, Parsons R, BadenTillson H, Pfannkoch C, Rogers YH, Smith HO. Environmental genome shotgun sequencing of the Sargasso sea. Science. 2004;304(5667):66–77. https://doi.org/10.1126/science.1093857.
 24.
Tyson GW, Chapman J, Hugenholtz P, Allen EE, Ram RJ, Richardson PM, Solovyev VV, Rubin EM, Rokhsar DS, Banfield JF. Community structure and metabolism through reconstruction of microbial genomes from the environment. Nature. 2004;428(6978):37–43.
 25.
Qin J, Li R, Raes J, Arumugam M, Burgdorf KS, Manichanh C, Nielsen T, Pons N, Levenez F, Yamada T, Mende DR, Li J, Xu J, Li S, Li D, Cao J, Wang B, Liang H, Zheng H, Xie Y, Tap J, Lepage P, Bertalan M, Batto JM, Hansen T, Le Paslier D, Linneberg A, Nielsen HB, Pelletier E, Renault P, SicheritzPonten T, Turner K, Zhu H, Yu C, Li S, Jian M, Zhou Y, Li Y, Zhang X, Li S, Qin N, Yang H, Wang J, Brunak S, Dore J, Guarner F, Kristiansen K, Pedersen O, Parkhill J, Weissenbach J, Bork P, Ehrlich SD, Wang J. A human gut microbial gene catalogue established by metagenomic sequencing. Nature. 2010;464(7285):59–65.
 26.
Veiga P, Gallini CA, Beal C, Michaud M, Delaney ML, DuBois A, Khlebnikov A, van Hylckama Vlieg JET, Punit S, Glickman JN, Onderdonk A, Glimcher LH, Garrett WS. Bifidobacterium animalis subsp. lactis fermented milk product reduces inflammation by altering a niche for colitogenic microbes. Proc Nat Acad Sci. 2010;107(42):18132–7. https://doi.org/10.1073/pnas.1011737107.
 27.
Turnbaugh PJ, Hamady M, Yatsunenko T, Cantarel BL, Duncan A, Ley RE, Sogin ML, Jones WJ, Roe BA, Affourtit JP, Egholm M, Henrissat B, Heath AC, Knight R, Gordon JI. A core gut microbiome in obese and lean twins. Nature. 2009;457(7728):480–4. https://doi.org/10.1038/nature07540.
 28.
Namiki T, Hachiya T, Tanaka H, Sakakibara Y. Metavelvet: an extension of velvet assembler to de novo metagenome assembly from short sequence reads. Nucleic Acids Res. 2012;40(20):155. https://doi.org/10.1093/nar/gks678.
 29.
Laserson J, Jojic V, Koller D. Genovo: de novo assembly for metagenomes. J Comput Biol. 2011;18(3):429–33. https://doi.org/10.1089/cmb.2010.0244.
 30.
Peng Y, Leung HCM, Yiu SM, Chin FYL. Metaidba: a de novo assembler for metagenomic data. Bioinformatics. 2011;27(13):94–101. https://doi.org/10.1093/bioinformatics/btr216.
 31.
Koren S, Treangen TJ, Pop M. Bambus 2: scaffolding metagenomes. Bioinformatics. 2011;27(21):2964–71. https://doi.org/10.1093/bioinformatics/btr520.
 32.
Peng Y, Leung HCM, Yiu SM, Chin FYL. Idbaud: a de novo assembler for singlecell and metagenomic sequencing data with highly uneven depth. Bioinformatics. 2012;28(11):1420–8. https://doi.org/10.1093/bioinformatics/bts174.
 33.
Boisvert S, Raymond F, Godzaridis É, Laviolette F, Corbeil J, et al. Ray Meta: scalable de novo metagenome assembly and profiling. Genome Biol. 2012;13(12):122.
 34.
Haider B, Ahn TH, Bushnell B, Chai J, Copeland A, Pan C. Omega: an overlapgraph de novo assembler for metagenomics. Bioinformatics. 2014;30(19):2717–22. https://doi.org/10.1093/bioinformatics/btu395.
 35.
Vingron M. Nearoptimal sequence alignment. Curr Opin Struct Biol. 1996;6(3):346–52.
 36.
Eppstein D. \(k\)best enumeration. Encyclopedia of algorithms. Berlin: Springer; 2015.
 37.
Vingron M, Argos P. Determination of reliable regions in protein sequence alignments. Protein Eng. 1990;3(7):565–9. https://doi.org/10.1093/protein/3.7.565.
 38.
Chao KM, et al. Locating wellconserved regions within a pairwise alignment. Comput Appl Biosci. 1993;9(4):387–96.
 39.
Costa MC. Persistency in maximum cardinality bipartite matchings. Oper Res Lett. 1994;15(3):143–9. https://doi.org/10.1016/01676377(94)900493.
 40.
Cechlárová K. Persistency in the assignment and transportation problems. Math Methods Oper Res. 1998;47(2):243–54. https://doi.org/10.1007/BF01194399.
 41.
Boros E, Golumbic MC, Levit VE. On the number of vertices belonging to all maximum stable sets of a graph. Discret Appl Math. 2002;124(1—3):17–25. https://doi.org/10.1016/S0166218X(01)003274.
 42.
Lacko V. Persistency in the traveling salesman problem on halin graphs. Discussiones Mathematicae Graph Theory. 2000;20(2):231–42. https://doi.org/10.7151/dmgt.1122.
 43.
Zenklusen R, Ries B, Picouleau C, de Werra D, Costa M, Bentz C. Blockers and transversals. Discret Math. 2009;309(13):4306–14. https://doi.org/10.1016/j.disc.2009.01.006.
 44.
Costa M, de Werra D, Picouleau C. Minimum dblockers and dtransversals in graphs. J Comb Optim. 2011;22(4):857–62. https://doi.org/10.1007/s1087801093346.
 45.
Pajouh FM, Boginski V, Pasiliao EL. Minimum vertex blocker clique problem. Networks. 2014;64(1):48–64. https://doi.org/10.1002/net.21556.
 46.
Italiano GF, Laura L, Santaroni F. Finding strong bridges and strong articulation points in linear time. Theor Comput. 2012;447:74–84. https://doi.org/10.1016/j.tcs.2011.11.011.
 47.
Farach M. Optimal suffix tree construction with large alphabets. In: Proc. 38th IEEE symposium on foundations of computer science (FOCS). 1997. p. 137–43.
 48.
Crochemore M, Rytter W. Jewels of stringology. Singapore: World Scientific Publishing; 2002. p. 1310.
 49.
Schrijver A. Combinatorial optimization. Berlin: Springer; 2003.
 50.
Mäkinen V, Belazzougui D, Cunial F, Tomescu AI. Genomescale algorithm design. Cambridge: Cambridge University Press; 2015.
 51.
Gabow HN, Tarjan RE. Faster scaling algorithms for network problems. SIAM J Comput. 1989;18(5):1013–36.
Authors' contributions
NOA and AIT devised the problem formulation, NOA found the graphtheoretic characterizations of safe walks, and NOA, VM and AIT devised the algorithms. All authors read and approved the final manuscript.
Acknowledgements
We thank Paul Medvedev for discussions on the proof of Theorem 4, and Martin Milanič for pointing us to persistent solutions and blockers.
Competing interests
The authors declare that they have no competing interests.
Availability of data and materials
Not applicable.
Consent for publication
Not applicable.
Ethics approval and consent to participate
Not applicable.
Funding
This work was partially supported by the Academy of Finland under Grant 284598 (CoECGR) to NOA and VM and Grant 274977 to AIT.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Author information
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
About this article
Received
Accepted
Published
DOI
Keywords
 Genome assembly
 Contig assembly
 Metagenomics
 Graph algorithm
 Circular walk