Posted 14 November 2009 - 06:29 AM
The GC average, of course, doesn't take into account local variation, but we've found the high average GC flank crosses in (and out) much more readily. This becomes an issue when trying to do allelic replacement, as you need the plasmid to cross out via the flank opposite to that used to cross in, or you wind up with only wild-type revertants. So, to balance the cross-in/cross-out favorability a bit, we try to balance the average GC of the flanks by shortening the high GC flank and/or lengthening the low GC flank until they're as close as we can get them in terms of average GC. Usually, we do this in increments of 200 - 500 bases, until we find a pair of flanks that match up as best we can, trying not to drop below 1 kb for either flank.
If the GC content difference is still around 3% or greater, we'll add a bit more (200 - 300 bp) to the low GC flank. We do most of our flank generation by PCR, so the final flank size wiggle is dictated by the primer design -- we use Primer3 to design primers, and take the best set possible based on our desired flank sizes.
This is really more of an art than a science, but by following these design guidelines, we've had pretty good success in such experiments, given the dearth of genetic tools available for Bacteroides.