Formula to determine coverage

[Tom Knight]

> Quite a qhile I remember seeing a formula that determines the minimum
> number of sub clones required of a certain size that should, within a
> certain confidence limit, give complete coverage of an insert of given
> length. It involved natural logs and things like that.

> Michael Mitchell                 "Smoke me a kipper,
> User Support                  I'll be back for breakfast."
> Molecular Biology Software      Ace Rimmer, Test Pilot
> +44 (0)171 269 3115                BBC-TV Red Dwarf           ENFJ

To get a desired probability P of having a section of your genome in
the library, where the library consists of clones each having a
fraction f of the original genome, you need a number of clones

       ln (1-P)
N =  ------------
       ln (1-f)

So, to get a 99% probability ( a common choice )

                    ln (1-.99)
N =  ------------------------------------------------
       ln (1 - (<size of insert> / <size of genome>))