Kudoa genomes from contaminated hosts reveal extensive gene order conservation and rapid sequence evolution

Myxozoans are obligate endoparasites that belong to the phylum Cnidaria . Compared to their closest free-living relatives, they have evolved highly simplified body plans and reduced genomes. Kudoa iwatai , for example, has lost upwards of two thirds of genes thought to have been present in its ancestors. However, little is known about myxozoan genome architecture because of a lack of sufficiently contiguous genome assemblies.

This work presents two new, near-chromosomal Kudoa genomes, built entirely from low-coverage long reads from infected fish samples. The results illustrate the potential of using unsupervised learning methods to disentangle sequences from different sources, and facilitate producing genomes from undersampled taxa. Extracting distinct components of chromatin interaction networks allows scaffolds from mixed samples to be assigned to their source genomes. Meanwhile, low-dimensional embeddings of read composition permit targeted assembly of potential parasite reads.

Despite drastic changes in genome architecture in the lineage leading to Kudoa and considerable sequence divergence between the two genomes, gene order is highly conserved. Although parasitic cnidarians show rapid protein evolution compared to their free-living relatives, there is limited evidence of less efficient selection. While deleterious substitutions may become fixed at a higher rate, large evolutionary distances between species make robustly analysing patterns of molecular evolution challenging. These observations highlight the importance of filling in taxonomic gaps, to allow a comprehensive assessment of the impacts of parasitism on genome evolution.