Two different datasets (gathered with the same process, but just different plots) look so similar that it seems likely that this 16nm value for the width of the glycosylation peak will be appropriate for the training plots to examine the other 80 dodecamers of SP-D for which i have pictures. The valus for all plots together (not divided into the four dodecamers) for old and new datasets below.  Please note that the red box at the bottom is an indication of deviation from a normal distribution of the dataset, this shows a significant skew which appears because of the variations in the way signal processing programs divide up the plot into peaks, and how I arrange the peaks into a number (15) that was determined by those same image and signal processing algorithms.

Here are additional, and hopefully the last, set of numbers for Peak 2 (tiny peak beside the N term peak) of SP-D, in terms of peak width.  I have used all numbers, including those which do not detect the peak, generated by plots of the trimers (as part of a plot for a hexamer, one CRD to the other CRD to determine that particular peak width.  I personally think that adding the missing values is a bit nuts but here are the data, including a new set of numbers which indicates that the peak is a non normal distribution.

It is fair to say that the peak width (when measured only on the values present (please note that the number of actual counted peaks is in parenthesis by N) is N=4, x=3.75+/ 0.34nm, which is consistent with measurements from the beginning of this study. For one dodecamer (42a_aka_44 see below on right hand side) this tiny peak was detected in more than half of the trimers plotted).

Four dodecamers of surfactant protein D, replicate tiny peak width (peak 2) which is a peak that shows up about 30% of the time. I have determined the peak width in nm in every which way i could, except combining the first (old) and second (new) datasets, which I chose not to do because there were some duplicates. The first dataset had every plot for every trimer that had been plotted up till that time. The second (new) dataset did not include the Gwyddion (red only in RGB) images, and so that left out all the limit range image filter images. The second dataset also did NOT include some of the far out filters in programs that were clearly not that usefor for determining structure. It is a select dataset that has a reasonable number of plots made mostly with images filtered with a gaussian blur, or median filter, and then peaks determined in 5 different signal processing programs (mentioned numerous times before, but including Octave (autofindpeaksplot, ipeak), Scipy, LagThresholdInfluence, and an excel template called PeakandValleyDetectionTemplate.

Note that the N termini junction appears twice (as there was no good way to divide the width of that peak in the trimers, as it represents a central combined value of a hexamer. Thus, a complete N term peak width is part of every trimer and appears as a complete peak each time.  I am actually happy that the combined mean nm width for the tiny peak is about what I had hoped, that is,  just around 3 nm.   30.1% is the occurrence from the first dataset 28% is the occurrence from the limited (second) dataset.  The mean of 2.9nm +/ 1.8nm is a summar from data with missing values, without missing values, first and second datasets, and those two criteria applied to each of the four trial dodecamers (as trimers) individually.  This works out to 16 separate tiny peak (peak 2) widths, for which the mean was found. Actual values are left hand column.

SEE NEXT entry for current numbers !!!