Impact of giving genome size instead of estimating it from reads ?

[RxLoutre]

Hello !

I am facing the same issues as mentioned here #133 . As I work with only one species of bacteria, their genome size is always between 6M and 10M. I was toying with the idea to speed up shovill process by giving a sort of "mean" expected genome size to speed up the process and avoid kmc freezes. But what would be the impact if for example I ask for an expected genome size of 7,5 and the genome size is in reality smaller than that ? Would it try to inflate the underlying assembly or not ? How is used this estimated genome size and how far from the actual size the estimation can be ?

Thank you for your help !

Best,