Phylogenetic signal from rearrangements in 18 Anopheles species by joint scaffolding extant and ancestral genomes

Abstract

Background: Genomes rearrangements carry valuable information for phylogenetic inference or the elucidation of molecular mechanisms of adaptation. However, the detection of genome rearrangements is often hampered by current deficiencies in data and methods: Genomes obtained from short sequence reads have generally very fragmented assemblies, and comparing multiple gene orders generally leads to computationally intractable algorithmic questions.

Results: We present a computational method, ADSEQ, which, by combining ancestral gene order reconstruction, comparative scaffolding and de novo scaffolding methods, overcomes these two caveats. ADSEQ provides simultaneously improved assemblies and ancestral genomes, with statistical supports on all local features. Compared to previous comparative methods, it runs in polynomial time, it samples solutions in a probabilistic space, and it can handle a significantly larger gene complement from the considered extant genomes, with complex histories including gene duplications and losses. We use ADSEQ to provide improved assemblies and a genome history made of duplications, losses, gene translocations, rearrangements, of 18 complete Anopheles genomes, including several important malaria vectors. We also provide additional support for a differentiated mode of evolution of the sex chromosome and of the autosomes in these mosquito genomes.

Conclusions: We demonstrate the method's ability to improve extant assemblies accurately through a procedure simulating realistic assembly fragmentation. We study a debated issue regarding the phylogeny of the Gambiae complex group of Anopheles genomes in the light of the evolution of chromosomal rearrangements, suggesting that the phylogenetic signal they carry can differ from the phylogenetic signal carried by gene sequences, more prone to introgression.

Keywords: Comparative genomics; Mosquito genomics; Scaffolding.

Fig. 1

Input and output of ADSEQ. (Left) Input data: (1) a species tree with extant genomes (A, B and C) containing observed adjacencies (black link) and scaffolding gene adjacencies with a prior score (blue link); each grey box represents a gene. (2) reconciled gene trees representing evolutionary histories of gene families annotated by evolutionary events. (Right) Typical output: gene order across ancestral and extant genomes including new extant gene adjacencies with a posterior score (green link) between genes located at fragments extremities in the initial genome assemblies

Fig. 2

Precision and recall statistics for scaffolding adjacencies on three artificially fragmented genomes (A.alb: A. albimanus, A.ara: A. arabiensis and A.dir: A. dirus). Left graph: results with 50% of reads. Right graph: results with all reads. The different methods results are plotted with the precision on the Y axis and the recall on the X axis. For ADSEQ and AD, results for three different adjacency support thresholds (0.1, 0.5 and 0.8) before genome linearization are plotted and represented with a color gradient. Note: A True Positive (TP) adjacency requires the proper orientation of both genes

Fig. 3

A. species trees (left: X phylogeny, right: WG phylogeny) with rearrangements per adjacency as branch lengths (× 10⁻³). The pie chart for a given species represents the adjacency degree of the genes of this species: orange represents genes with no adjacency, light blue genes with adjacency degree of 1 and green genes with adjacency degree of 2. Moreover, the diameter of each chart is proportional to the number of genes in the corresponding species

Fig. 4

Left: Number of segments in extant and ancestral genomes, according to three runs of ADSEQ in three different conditions. In the first run, we turn off the scaffolding mode on the X phylogeny, that is, it only constructs ancestral segments. The first column “XNS ext” thus describes the initial assembly, and “XNS anc” the assembly of ancestral genomes when reconstructed without extant scaffolding. In the second and third runs, the scaffolding mode was turned on, and run with the X phylogeny (“X ext” and “X anc”) and the WG phylogeny (“WG ext” and “WG anc”). Right: Number of rearrangements over all branches of the X phylogeny, with and without the scaffolding mode