Nick Oswald at Bitesize Bio considers a study of various approaches to desalting DNA for electroporation and concludes that "for the high concentration samples the microcolumns were the clear winner, although dialysis and gel filtration gave just a 2-fold lower efficiency." He adds that microcolumns were also best for low concentration samples -- and in this case were "at least 100x more efficient than any of the other methods." On a side note, he says, "For some reason, ethanol precipitation, a widely used method of desalting DNA, gave miserable results for both samples."