The accuracy of Apis mellifera reference genomes: from version Amel 3.0 to Amel 4.5

The availabilty of reference genomes is accelerating rapidly, making their use in a wide variety of biological research programmes more feasible than ever. However, current Next-Generation Sequencing platforms are limited in the length of reads they are able to produce; requiring the correct order to be determined algorithmically. While there is a potential for errors in assembly algorithims, genetic pedigree data can be used to identify recombination events and, as recombination events are rare locally, test the order of sequences within a genome assembly. We use high-resolution population genomic data to test and compare the assembly quality of the three most recent reference genome assemblies for the western honey bee (Apis mellifera). As a model organism, there are several reference genomes available for A. mellifera with estimated recombination rates ranging from 19 cM/Mb to 37 cM/Mb. We identify a large degree of variation between assemblies and find that at least 20% of the most recent A. mellifera reference genome is mis-assembled. Providing an explanation for the degree of variation in estimated recombination rates and potentially influencing results downstream.