The minimum path cover problem asks us to find a minimum-cardinality set of paths that cover all the nodes of a directed acyclic graph (DAG). We study the case when the size k of a minimum path cover is small, that is, when the DAG has a small width . This case is motivated by applications in pan-genomics , where the genomic variation of a population is expressed as a DAG. We observe that classical alignment algorithms exploiting sparse dynamic programming can be extended to the sequence-against-DAG case by mimicking the algorithm for sequences on each path of a minimum path cover and handling an evaluation order anomaly with reachability queries . Namely, we introduce a general framework for DAG-extensions of sparse dynamic programming. This framework produces algorithms that are slower than their counterparts on sequences only by a factor k . We illustrate this on two classical problems extended to DAGs: longest increasing subsequence and longest common subsequence . For the former, we obtain an algorithm with running time O ( k | E |log | V |). This matches the optimal solution to the classical problem variant when the input sequence is modeled as a path. We obtain an analogous result for the longest common subsequence problem. We then apply this technique to the co-linear chaining problem, which is a generalization of the above two problems. The algorithm for this problem turns out to be more involved, needing further ingredients, such as an FM-index tailored for large alphabets and a two-dimensional range search tree modified to support range maximum queries. We also study a general sequence-to-DAG alignment formulation that allows affine gap costs in the sequence. The main ingredient of the proposed framework is a new algorithm for finding a minimum path cover of a DAG ( V , E ) in O ( k | E |log | V |) time, improving all known time-bounds when k is small and the DAG is not too dense. In addition to boosting the sparse dynamic programming framework, an immediate consequence of this new minimum path cover algorithm is an improved space/time tradeoff for reachability queries in arbitrary directed graphs.

Transcript prediction can be modeled as a graph problem where exons are modeled as nodes and reads spanning two or more exons are modeled as exon chains. Pacific Biosciences third-generation sequencing technology produces significantly longer reads than earlier second-generation sequencing technologies, which gives valuable information about longer exon chains in a graph. However, with the high error rates of third-generation sequencing, aligning long reads correctly around the splice sites is a challenging task. Incorrect alignments lead to spurious nodes and arcs in the graph, which in turn lead to incorrect transcript predictions. We survey several approaches to find the exon chains corresponding to long reads in a splicing graph, and experimentally study the performance of these methods using simulated data to allow for sensitivity/precision analysis. Our experiments show that short reads from second-generation sequencing can be used to significantly improve exon chain correctness either by error-correcting the long reads before splicing graph creation, or by using them to create a splicing graph on which the long-read alignments are then projected. We also study the memory and time consumption of various modules, and show that accurate exon chains lead to significantly increased transcript prediction accuracy. AVAILABILITY: The simulated data and in-house scripts used for this article are available at http://www.cs.helsinki.fi/group/gsa/exon-chains/exon-chains-bib.tar.bz2.

RNA-Seq technology offers new high-throughput ways for transcript identification and quantification based on short reads, and has recently attracted great interest. This is achieved by constructing a weighted DAG whose vertices stand for exons, and whose arcs stand for split alignments of the RNA-Seq reads to the exons. The task consists of finding a number of paths, together with their expression levels, which optimally explain the weights of the graph under various fitting functions, such as least sum of squared residuals. In (Tomescu et al. BMC Bioinformatics, 2013) we studied this genome-guided multi-assembly problem when the number of allowed solution paths was linear in the number of arcs. In this paper, we further refine this problem by asking for a bounded number k of solution paths, which is the setting of most practical interest. We formulate this problem in very broad terms, and show that for many choices of the fitting function it becomes NP-hard. Nevertheless, we identify a natural graph parameter of a DAG G, which we call arc-width and denote ⟨G⟩, and give a dynamic programming algorithm running in time O(W(k)⟨G⟩(k)(⟨G⟩+ k)n) , where n is the number of vertices and W is the maximum weight of G. This implies that the problem is fixed-parameter tractable (FPT) in the parameters W, ⟨G⟩, and k. We also show that the arc-width of DAGs constructed from simulated and real RNA-Seq reads is small in practice. Finally, we study the approximability of this problem, and, in particular, give a fully polynomial-time approximation scheme (FPTAS) for the case when the fitting function penalizes the maximum ratio between the weights of the arcs and their predicted coverage.

BACKGROUND: Through transcription and alternative splicing, a gene can be transcribed into different RNA sequences (isoforms), depending on the individual, on the tissue the cell is in, or in response to some stimuli. Recent RNA-Seq technology allows for new high-throughput ways for isoform identification and quantification based on short reads, and various methods have been put forward for this non-trivial problem. RESULTS: In this paper we propose a novel radically different method based on minimum-cost network flows. This has a two-fold advantage: on the one hand, it translates the problem as an established one in the field of network flows, which can be solved in polynomial time, with different existing solvers; on the other hand, it is general enough to encompass many of the previous proposals under the least sum of squares model. Our method works as follows: in order to find the transcripts which best explain, under a given fitness model, a splicing graph resulting from an RNA-Seq experiment, we find a min-cost flow in an offset flow network, under an equivalent cost model. Under very weak assumptions on the fitness model, the optimal flow can be computed in polynomial time. Parsimoniously splitting the flow back into few path transcripts can be done with any of the heuristics and approximations available from the theory of network flows. In the present implementation, we choose the simple strategy of repeatedly removing the heaviest path. CONCLUSIONS: We proposed a new very general method based on network flows for a multiassembly problem arising from isoform identification and quantification with RNA-Seq. Experimental results on prediction accuracy show that our method is very competitive with popular tools such as Cufflinks and IsoLasso. Our tool, called Traph (Transcrips in gRAPHs), is available at: http://www.cs.helsinki.fi/gsa/traph/.