PCR ghost bands

[Pierre Jelenc]

Very often (3/4 of the time), I see a PCR product shorter than the correct
one. This extra band is sometimes much stronger than the correct one, and
the difference in size between correct and incorrect is pretty much random
from primer pair to primer pair. I.e. it is not a fixed number of
nucleotides, nor is it proportional to the size of the correct fragment. 

This extra band is _always_ shorter, typically by 10-30 nucleotides, and 
it is absolutely reproducible with a given primer pair, regardless of PCR 
conditions (Mg concentration, annealing temperature, elongation time, 
number of cycles between 20 and 40, etc). It is readily visible in 
denaturing acrylamide gels, and seems to be there in concentrated agarose 
gels as well, although the lower resolution in the latter makes it 
difficult to be sure.

It does not appear to be a gel artifact, because I have run side by side 
reactions with and without extra bands, nor to be simply due to the 
reaction conditions, because I run several reactions from the same premix 
of reagents, added to the template and primers that have been premelted 
and annealed under wax.

I am at a complete loss.

Pierre

Pierre Jelenc                        pcj1 at columbia.edu 
                                    Columbia University, New York