Bins and Binnning Definitions
Previous Top Next

There is some confusion about the terms 'bin' and 'superbin' in the literature. Therefore, we define the terms as follows:
  
Bin
A bin is equivalent to a peak in a trace curve or a band in a gel image. It holds data values around a local maximum of a single sample. It has a certain width in data points and after sizing also a width in base sizes.

It comprises data points with the highest data points usually forming a local maximum near the center. The sum of all data point values is the area.

The base size of a bin should be an integer value because DNA chains can only deviate from each other by the addition or deletion of a single nucleotide. In reality the base size is a float number. This is due to the inaccurateness of biological data.

Bins that below the smallest or beyond the highest size standard fragment should not be considered for further analysis because they cannot be sized exactly (a limitation of the local southern algorithm).
 
graphic
Superbin
Sometimes two or more bins of the same sample (trace peaks/image bands) are too close to each other to form individual bins. In this case the individual bins are united in a superbin. So a superbin is a container for bins located closely together.

An example:
A sample has two peaks sized with 79.6 and 80.2. They are 0.6 bases apart from each other. Because DNA fragments can differ in a minimum of a single base in length, these peaks may not be treated as separate fragments. If the values are rounded we obtain 80 for both. So a clear decision is difficult. Therefore we define a superbin comprising the two bins.
 
graphic
Hyperbin
A hyperbin is a container for bins and superbins of two or more samples.

If several DNA fingerprints are compared it is a good idea to check the presence of certain peaks in different samples.

Example
One sample has a peak with a size of 83 base pairs ('peak at 83 bp'). This peak is also present in the second sample but not in the third. The second and third sample have a common peak with a size of 108 base pairs. And so on. This can be written as a table:
  
 
 
Peak at 83 bp
Peak at 108 bp
...
Sample #1
x
 
...
Sample #2
x
x
...
Sample #3
 
x
...
...
...
...
...

So the peaks of individual samples are united in larger containers, the hyperbins.

How can we build hyperbins?
The basic idea would be to cluster individual sample bins that vary in a range of ±0.5 from a defined base size into the same hyperbin. This hyperbin would have a width of exactly 1.0.

In reallity it is much more complicated...
graphic