Question: We used Supernova to assemble genomes of individuals from the same population with similar heterozygosity, but slightly different molecule size (e.g. 45 kb vs. 55 kb). It appears that there is no direct correlation between molecule size and assembly quality (for example, scaffold N50). What can explain this?
Answer: In this case, the difference between a mean of 45 kb and 55 kb is not incredibly large, and other factors also impact assembly quality. Sequencing quality, coverage depth, and differences between individuals in a population all influence the final outcome. The length-weighted mean molecule length is a summary statistic that assumes molecule length is exponentially distributed, but there could be deviations in this distribution.
For more information please see Supernova Performance.