CSA: A high-throughput chromosome-scale assembly pipeline for vertebrate genomes

BACKGROUND: Easy-to-use and fast bioinformatics pipelines for long-read assembly that go beyond the contig level to generate highly continuous chromosome-scale genomes from raw data remain scarce. RESULT: Chromosome-Scale Assembler (CSA) is a novel computationally highly efficient bioinformatics pipeline that fills this gap. CSA integrates information from scaffolded assemblies (e.g., Hi-C or 10X Genomics) or even from diverged reference genomes into the assembly process. As CSA performs automated assembly of chromosome-sized scaffolds, we benchmark its performance against state-of-the-art reference genomes, i.e., conventionally built in a laborious fashion using multiple separate assembly tools and manual curation. CSA increases the contig lengths using scaffolding, local re-assembly, and gap closing. On certain datasets, initial contig N50 may be increased up to 4.5-fold. For smaller vertebrate genomes, chromosome-scale assemblies can be achieved within 12 h using low-cost, high-end desktop computers. Mammalian genomes can be processed within 16 h on compute-servers. Using diverged reference genomes for fish, birds, and mammals, we demonstrate that CSA calculates chromosome-scale assemblies from long-read data and genome comparisons alone. Even contig-level draft assemblies of diverged genomes are helpful for reconstructing chromosome-scale sequences. CSA is also capable of assembling ultra-long reads. CONCLUSIONS: CSA can speed up and simplify chromosome-level assembly and significantly lower costs of large-scale family-level vertebrate genome projects.