Hares (genus Lepus) provide clear examples of repeated and often massive introgressive hybridization and striking local adaptations. Genomic studies on this group have so far relied on comparisons to the European rabbit (Oryctolagus cuniculus) reference genome. Here, we report the first de novo draft reference genome for a hare species, the mountain hare (Lepus timidus), and evaluate the efficacy of whole-genome re-sequencing analyses using the new reference versus using the rabbit reference genome. The genome was assembled using the ALLPATHS-LG protocol with a combination of overlapping pair and mate-pair Illumina sequencing (77x coverage). The assembly contained 32,294 scaffolds with a total length of 2.7 Gb and a scaffold N50 of 3.4 Mb. Re-scaffolding based on the rabbit reference reduced the total number of scaffolds to 4,205 with a scaffold N50 of 194 Mb. A correspondence was found between 22 of these hare scaffolds and the rabbit chromosomes, based on gene content and direct alignment. We annotated 24,578 protein coding genes by combining ab-initio predictions, homology search, and transcriptome data, of which 683 were solely derived from hare-specific transcriptome data. The hare reference genome is therefore a new resource to discover and investigate hare-specific variation. Similar estimates of heterozygosity and inferred demographic history profiles were obtained when mapping hare whole-genome re-sequencing data to the new hare draft genome or to alternative references based on the rabbit genome. Our results validate previous reference-based strategies and suggest that the chromosome-scale hare draft genome should enable chromosome-wide analyses and genome scans on hares.

DNA Sampling, Extraction, and Sequencing

One female mountain hare (Lepus timidus hibernicus) specimen (NCBI BioSample ID SAMN12621015) was captured from the wild for scientific research purposes by the Irish Coursing Club (ICC) at Borris-in-Ossory, County Laois under National Parks & Wildlife (NPWS) license no. C 337/2012 issued by the Department of Arts, Heritage and the Gaeltacht (dated October 31, 2012). Genomic DNA was extracted from kidney, muscle, and ear tissue using the JETquick Tissue DNA Spin Kit (GENOMED), with RNAse and proteinase K to remove RNA and protein contamination. Genomic libraries of different insert lengths were generated following the standard ALLPATHS-LG protocol (Gnerre et al. 2011): one Illumina TruSeq DNA library of 180 bp fragments was sequenced with overlapping paired-end (OPE) reads, and three Illumina TruSeq DNA mate-pair (MP) libraries of 2.5, 4.5, and 8.0 kb insert sizes. Whole-genome sequencing was performed at The Genome Analysis Center (TGAC, currently Earlham Institute, Norwich, UK)—seven HiSeq2000 lanes (five OPE and two 4.5 kb MP)—and CIBIO’s New-Gen sequencing platform—three HiSeq1500 lanes (2.5 and 8.0 kb MP). Raw sequencing reads were deposited in the Sequence Read Archive.

Genome Assembly

De novo assembly was performed using ALLPATHS-LG (Gnerre et al. 2011) with default parameters using OPE and mate-pair reads. The resulting assembly was evaluated with REAPR v1.0.18 (Hunt et al. 2013) to break incorrect scaffolds, by mapping the paired-end and the 4.5 kb mate-pair reads on the assembled genome. Another round of scaffolding was then performed using SSPACE v3.0 (Boetzer et al. 2011), with a minimum overlap of 32 bp and supported by a minimum of 20 reads (CIBIO-ISEM_LeTim1.0_Assembled.fasta.gz).

Finally, we leveraged the existence of the high-quality assembly of the genome of the European rabbit (Oryctolagus cuniculus—Ensembl OryCun2.0), to improve the contiguity of the assembly using the reference-based scaffolder MeDuSa v.1.6 with five iterations (Bosi et al. 2015) (CIBIO-ISEM_LeTim1.0_re-scaffolded.fasta.gz).

This re-scaffolding orders and re-orientates scaffolds without affecting intra-scaffold sequence. Quality control of the assembly at different stages was assessed based on metrics obtained with QUAST v.3.2 (Mikheenko et al. 2016). The completeness of the L. timidus re-scaffolded genome was evaluated using BUSCO v.3.0.2 (Simão et al. 2015), based on the presence and absence of core single-copy genes (from mammalia_odb9 database). We then checked consistency of gene content in the larger chromosome-like scaffolds and rabbit chromosomes using blastp from NCBI BLAST v2.7.1+ (Camacho et al. 2009), considering the best hit per gene with similarity above 90% over 500 bp. The 22 rabbit chromosomes were aligned against inferred corresponding L. timidus re-scaffolded scaffolds using D-Genies v. 1.2.0 Mashmap (Cabanettes and Klopp 2018).

Genome Annotation

Repetitive regions were identified using RepeatModeler v.1.0.11 (Smit and Hubley 2008) and masked using RepeatMasker v.4.0.7 (Smit et al. 2013). The masked genome was used as input for gene prediction in MAKER v.3.01.02 (Cantarel et al. 2008), using ab-initio predictions, L. timidus transcriptome data, and rabbit protein annotations (O. cuniculus) - (CIBIO-ISEM_LeTim1.0.cdna.abinitio.fa.gz and CIBIO-ISEM_LeTim1.0.pep.abinitio.fa.gz). Functional inference for genes and transcripts was performed using the translated CDS features of each coding transcript. Each predicted protein sequence was based on blastp searches against the Uniprot-Swissprot database to retrieve gene name and function, and InterProscan v5.30-69 (Jones et al. 2014) to retrieve Interpro, Pfam v31.0 (Finn et al. 2016), GO (Mi et al. 2017), KEGG (Kanehisa et al. 2016), and Reactome (Fabregat et al. 2018) information (annotation files: CIBIO-ISEM_LeTim1.0_re-scaffolded.gff.gz and CIBIO-ISEM_LeTim1.0_re-scaffolded.id.map.gz).

This assembly and annotation files corresponds to the version CIBIO-ISEM_LeTim1.0 reported in the article "An Annotated Draft Genome of the Mountain Hare (Lepus timidus)"