Co-linear Chaining on Pangenome Graphs

Abstract

Pangenome reference graphs are useful in genomics because they compactly represent the genetic diversity within a species, a capability that linear references lack. However, efficiently aligning sequences to these graphs with complex topology and cycles can be challenging. The seed-chain-extend based alignment algorithms use co-linear chaining as a standard technique to identify a good cluster of exact seed matches that can be combined to form an alignment. Recent works show how the co-linear chaining problem can be efficiently solved for acyclic pangenome graphs by exploiting their small width [Makinen et al., TALG'19] and how incorporating gap cost in the scoring function improves alignment accuracy [Chandra and Jain, RECOMB'23]. However, it remains open on how to effectively generalize these techniques for general pangenome graphs which contain cycles. Here we present the first practical formulation and an exact algorithm for co-linear chaining on cyclic pangenome graphs. We rigorously prove the correctness and computational complexity of the proposed algorithm. We evaluate the empirical performance of our algorithm by aligning simulated long reads from the human genome to a cyclic pangenome graph constructed from 95 publicly available haplotype-resolved human genome assemblies. While the existing heuristic-based algorithms are faster, the proposed algorithm provides a significant advantage in terms of accuracy. Implementation: https://github.com/at-cg/PanAligner

Haplotype-aware sequence alignment to pangenome graphs

“…Several versions of co-linear chaining problems have been studied for aligning two sequences [1,20,32,37,10,11,40]. Recent works have further studied the extension of chaining on acyclic [33,30,6,46] and cyclic pangenome graphs [3,43] but these formulations do not consider the haplotype paths. .y] in the DAG (Figure 3).…”

“…These formulations do not consider the associations between genetic variants and may lead to alignments with spurious recombinations in variant-dense regions of the graph [42]. The existing formulations for co-linear chaining on graphs share the same limitation [6,33,46,30,43]. Chaining on DAGs can be solved in O(KN log KN ) time, where K is the minimum number of paths covering all the vertices and N is the number of exact matches between the query and the DAG [6,30].…”

Label-guided seed-chain-extend alignment on annotated De Bruijn graphs

“…Some works determine a traversal distance between nodes on-the-fly by traversing local neighbourhoods around nodes [29,30]. More efficient strategies require a decomposition of the graph, typically into subgraphs [28,47] or a path/walk cover [31][32][33]46]. After chaining, anchor extension searches the graph forwards and backwards from the ends of each anchor to find high-scoring walks.…”

“…We pre-process the graph and generate walks using the procedure described by [33]. To reduce the final representation size, we refrain from maintaining a matrix of traversal distances from each node to each stored walk.…”

“…A key search task on these indexes is sequence-to-graph alignment, a generalisation of pairwise sequence-to-sequence alignment (i.e., computing the maximum similarity score between a query and a target sequence). In this setting, the target sequences are the spellings of contiguous walks ( §2.1) on the sequence graph [24][25][26][27][28][29][30][31][32][33][34]. Many of these tools follow a three-step seed-chain-extend search paradigm ( §2.2).…”