Using the same PCR plate in more runs - Just curious about it. (Sep/08/2010 )

We have a LightCycler 480, it uses 96-well plates and that's the only thing it can use (no strips, no single tubes).
Of course in some experiments we don't need 96 reactions in one run, often like only 10 or even less when optimizing. The plates are pretty expensive so we think why not to use the unused wells (no recyclation!) in a second/third/... run with another samples.

It's done like this: we take the finished plate and cut the transparent seal around the place where we gonna put our samples and remove it (we keep the PCR-ed wells covered to avoid contamination). It's better to cut at least one well on each side bigger space, than we need. We pipette the new reactions in and seal it with the new transparent cover cut only to overlap the opened space.
The edges of the overlap are near the wells one well apart from our samples, as we have experience that if the edge is near our samples, it can somehow screw the reading of the border well.

So far it should be no problem, all wells are covered and samples are in the wells, that were never used, they were only empty sealed in the first run. But actually what happens from the third run (to be exact in second run there are some minor differences, but unimportant) of the same plate, that the fluorescent readings are jumping up and down, even in the baseline region, it creates a jagged curve all the way. Specifity or sensitivity of the reaction is not affected, but of course if you have jagged exponential phase of the curve, the software has huge difficulties to compute the Ct.

I'm really curious of what could be the cause. It doesn't matter if the plate is "fresh" or been used last month. I was thinking about the temperature deformation, because original Roche plates are prone to that, but even when I used the 4titude plates that should minimise it, it still happens.

my guess is temperature changes to the plate as well. in particular the light might refract around in some weird way after the deformation. i'd be curious to know if it occurs using white plates (if available for the lightcycler)