best N50

[mmmm]

is there a reference for the range of N50 for a bacterial genome (5-6Mb)?- does 200kbp (N50) look ok or is it too high (as in Illumia tech note, it is mentioned that 80kbp was the best N50 for ecoli genome??

[Brian Bushnell]

Higher is always better. More repetitive genomes, and lower-quality or shorter reads will reduce the N50, but there's no reason to reduce it intentionally.

[kbradnam]

An N50 of 200 Kbp is better than 199 Kbp and worse than 201 Kbp. Beyond that, be careful about relying too much on N50.

High N50s can still be produced from genome assemblies which are terrible in many other ways. This is something I blogged about recently:

http://www.acgt.me/blog/2014/4/8/which-genome-assembler-gives-you-the-best-genome-assembly

Less of an issue for bacterial genomes is the fact that N50 also counts unknown bases. You can have 'long' scaffolds which are the result of scaffolding errors which insert thousands of unknown bases into a sequence.

My understanding of modern bacterial genome sequencing is that people are now reducing the whole genome down to just a few contigs (sometimes just 1). For bacteria, the number of contigs in your assembly is perhaps more revealing than the N50 length.