Karyn Meltz Steinberg, Richard Wilson and colleagues produced a single-haplotype assembly of the human genome using genomic DNA from an essentially haploid hydatidiform mole (Genome Res. 10.1101/gr.180893.114; 4 November 2014). They used a short read–based, reference-guided method to assemble the complete genome of the CHM1 hydatidiform mole, using 100× whole-genome shotgun sequence reads and the GRCh37 reference assembly. The assembly was then improved through the integration of high-quality finished sequence from 382 BAC clones, selected to overlap complex regions with segmental duplications or where the reference had known gaps. They demonstrate that the CHM1_1.1 assembly is of high quality, with a contig N50 length of 144 kb and a scaffold N50 length of 50 Mb, making it more contiguous than previous human whole-genome assemblies. Although CHM1_1.1 is of high quality, the authors note errors in it, including ones due to the use of the GRCh37 reference. They stress the value of using multiple sequencing technologies and assembly-independent resources, demonstrated by alignment to a more recently available long-read data set, which allowed the identification of specific assembly errors.