Quantitative analysis of reactivities.

Analysis of raw electropherograms, which contained data for fluorescence intensity versus elution time, was carried out as described previously (70). The data files were analyzed in ShapeFinder (37). Data from the 6-FAM and VIC channels for each of the four capillaries were first combined into a single file. A fitted baseline adjustment with a window width of 200 was applied to all eight channels. Cubic mobility shifts were performed manually on the four 6-FAM channels to align these identical ddATP sequencing reactions. The shift values from these alignments were then applied to the appropriate VIC channels, aligning the experimental (plus), control (minus), and sequencing channels with one another. The data from the four 6-FAM channels were then removed before continuing the analysis. Signal decay correction was applied to each of the four remaining VIC channels. The region of interest for the signal decay correction and further analysis was determined by visually inspecting the quality of peaks within the plus channel, and the same window was applied to all four channels. The rescale factor and equation parameters (A, q, and C) were kept at their default values of 10,000, 1,000,000, 0.999, and 10,000, respectively. A scale factor was then applied to the minus channel so that most peaks were of a height equal to or less than that of the corresponding plus peak. This factor was typically between 0.3 and 0.7. Alignment and integration were then performed over a range equal to or narrower than that used for signal decay correction. After manual inspection and correction of peaks, the ShapeFinder software performed a whole-channel Gaussian integration to quantify all individual peak areas, and the minus peak areas were subtracted from the plus areas. The raw reactivity data for each primer were normalized using a 2% to 8% normalization scheme (56). Accordingly, the top 2% of data were taken as outliers. The average of the next 8% of the data was found, and each datum, including the top 2%, was then divided by this average reactivity. At least three independent runs were completed for each primer. The normalized reactivity values for each 5′ UTR position were tested for outliers using Dixon’s Q test at a 95% confidence interval. The outliers were excluded from further analysis.