The problem of de-novo assembly for metagenomes using only long reads is gaining attention. We study whether post-processing metagenomic assemblies with the original input long reads can result in quality improvement. Previous approaches have focused on pre-processing reads and optimizing assemblers. |