Benchmarks

Short read assembly

JGI researchers participated in the first Assemblathon benchmark to evaluate progress in algorithms for de novo assembly of short reads. The Assemblathon team provided simulated Illumina datasets for a polymorphic mammalian chromosome publicly available, and assemblies were submitted by the genome assembly research community for evaluation and comparison. All major assembly groups participated. JGI's Meraculous method had the best structural accuracy and was effectively tied for first place in sequence accuracy and scaffold size. Overall the meraculous entry was ranked fourth overall (out of 20 entries). This first Assemblathon competition provided useful impartial feedback for the ongoing improvement of meraculous, resulting in a focus on improving contig length, completeness,and handling of copy number variation -- the three Assemblathon 1 metrics where meraculous was not among the top 4. The JGI group is actively involved in the second Assemblathon competition, which is focused on three polymorphic gigabase-sized vertebrate genomes, and will again be evaluated by the Assemblathon team relative to entries from other leading groups.

1. Results
2. References:
   • Meraculous
   • Assemblathon