Reference Genome Assembly in 2015 - Methods for the 21st Century

High quality reference genomes are the foundation for biological analyses.  In the last century, methods for creating high quality reference genomes relied upon scarce and expensive Sanger sequences with maps developed from libraries of large insert BAC or Fosmid clones.  Advances early in this century have significantly reduced the cost per base of sequence generation, making the raw Illumina sequence plentiful but shorter than Sanger data.  Alternative, long-read methods such as Pacific Biosciences can complement these short read data by filling in gaps and connecting contigs into larger scaffolds despite their higher error rates.  Additional newer methods such as optical maps (BioNano Genomics and OpGen) and Hi-C sequence foster the generation of much more contiguous scaffolds that complement the assembly methods that use short and long read data. 

Here we report on our experience and experiments combining these data to produce better genome reference sequences. Experiments with human data, where we have a finished reference genome, have informed our current preferred methods.  We are applying these methods to produce an improved reference genome for the Drosophila pseudoobscura.  Illumina assembly methods produced an excellent draft assembly (90 kb Contig N50, 2.3 Mb Scaffold N50) and our established PacBio with PBJelly method improves the Contig N50 multi-fold (295 kb).  Alternate strategies to error correct and de novo assemble the PacBio data using HGAP/Falcon produce 400 to 750 kb Contig N50.  We will present these data and progress toward a finished quality assembly.