SOE or Megaprimer PCR mutant libraries ?

[the End]

Has anyone sucessfully used the SOE (splicing by overlap extension) or
"megaprimer" methods to incorporate a mixed oligo into a cassette 
which replaces a region lacking convenient flanking restriction sites ?

I have only found one article (Biotechniques 14(3) 454) that mentions 
SOE in making a mutant library. The authors (Morrison and Desrosiers
at Harvard) found that the mutation frequency obtained after the 
incorporation of the primer mixture by SOE was somewhat less than
expected from the composition of the primers themselves.

Can anyone suggest a reason for lowered mutation frequency other than 
competiton among primers for annealing sites ? Has anyone tried the 
SOE procedure without adding the outer-most flanking primer to 
the second round of PCR ? This is in effect a "double megaprimer"
approach.

The megaprimer method would be most convenient for me (one less primer),
however, I see no reports of it in randomizing a mutagenesis target.

What concerns me about both procedures is there will be a competition 
between the "megaprimer" strands reanealing and priming the other 
target strand which will strongly favor reanealing. This then would 
generate a few full length targets inefficiently, which would then be 
amplified by the flanking primers (in the cae of SOE). In the three 
primer (megaprimer) method, this inefficient early product is not 
further amplified, and so the process is inherently inefficient
(Biotechniques 14(3) 366). Does it work ?

As I only need to introduce on average one change in a twenty base target,
or look at about 2000 clones, am I overly concerend about the representation 
of the oligos in the resulting product ?

Sorry I haven't mastered the ASCII PCR diagram yet.

Thanks much,

Jim
J. Graham