Chromosomal scale assembly of Cotesia congregata genome based on chromosomal contact mapping. Chromosome scale assembly of C. congregata genome Hi-C library preparation The individual wasps had their gut removed and were immediately suspended after sampling in 30 mL of 1X tris-EDTA buffer and formaldehyde at 3% concentration, then fixated for one hour. 10 mL of glycine at 2.5 M was added to the mix for quenching during 20 min. A centrifugation recovered the resulting pellet for −80°C storage and awaiting further use. The libraries were then prepared and sequenced (2 × 75 bp, paired-end Illumina NextSeq with the first ten bases acting as tags), as previously described57 using the DpnII enzyme. Read processing and Hi-C map generation The Hi-C read library was processed and mapped onto DpnII fragments of the reference assembly using HiCbox (available at https://github.com/kozullab/HiCbox) with bowtie258 on the back-end (option --very-sensitive-local, discarding alignments with mapping quality below 30). Fragments were then filtered according to size and coverage distribution, discarding sequences shorter than 50 bp or below one standard deviation away from the mean coverage. Both trimmed contact maps were then recursively sum-pooled four times by groups of three, yielding bins of 34= 81 fragments. Contact-based re-assembly The genome was re-assembled using an updated version of GRAAL (dubbed instaGRAAL) for large genomes on the aforementioned contact maps for 1,000 cycles, as described in59. Briefly, the program modifies the relative positions and/or orientations of sequences according to expected contacts given by a polymer model. These modifications take the form of a fixed set of operations (swapping, flipping, inserting, merging, etc.) on the 81-fragment bins. The likelihood of the model is maximizing by sampling the parameters with an MCMC method. After a number of iterations, the contact distribution converges and the global genome structure ceases to evolve, at which point the genome is considered reassembled. The process yielded eleven main scaffolds comprising >99% of the bin sequences. Polishing and post-assembly processing Each scaffold was independently polished by reconstructing the initial contig structure whenever relocations or inversions are found. In addition, previously filtered sequences were reintegrated next to their original neighbors in their initial contig when applicable. The implementation is part of instaGRAAL polishing and available at https://github.com/koszullab/instaGRAAL (run using the –polishing mode). Assembly validation We performed the validation with QUAST-LG60, an updated version of the QUAST analyzer tailored for large genomes. The initial assembly from Illumina short reads was used as reference. The assessed metrics include genomic feature completeness, Nx and NGx statistics as well as global and local misassemblies. In addition, each assembly was assessed for ortholog completeness with BUSCO v361. The reference dataset used for expected gene content was pulled from the OrthoDB (version 9) database for Hymenoptera, comprising 4,415 orthologs.