Marker Locking & normalisation

The position of the bands will not be the same on every gel. There are a number of reasons for this.

  • Experimental conditions may stretch one gel with respect to the other.
  • There may be local variations in migration rate.
  • The well may vary in location between gels.

In order to compare the bands from different gels is it necessary to normalize the band position. This is done by normalising bands on a particular gel to a standard gel. This normalisation is done in two stages. Firstly, a mapping is established between the bands in the marker lanes and the set of standard bands defined in the standard file this is the marker Locking stage. Secondly the mapping is extended to sample lanes by linear interpolation between marker lanes and applied, the band normalisation stage.

Marker locking

Typically the band caller will not automatically call the correct bands for marker locking. This can be for a variety of reasons.

  • Bands in the standard set may be missing from some marker lanes.
  • There may be extra fragments in some marker lanes.

before marker locking proceeds the marker lanes must be edited so that each marker lane has the correct number of bands. Once the number of bands in each marker lane is the same as the number of standard bands then a mapping can be established between position in the marker lane and position on the standard gel. A example of this mapping shown below

Band normalisation trace

Band normalisation trace

zoom

The upper part of the figure shows the central trace and the bands in the marker lanes. The triangles in the lower part of the plot corresponding to the bands in the standard lane.

Band normalisation

Using the data obtained in the previous step, bands in the sample lanes can be normalized. Linear interpolation is used between standard lanes to estimate the position of the standard bands in the sample lanes. Image produces migration distances for each band by interpolation between the standard bands and can optionally provide fragment size information if a standard is available. This information is transformed using a logarithmic scale for normalisation to ensure greater accuracy. The interpolation of migration distance is either linear or non-linear the choice is user configurable whereas the size interpolation is always non-linear.

Image can also provide a normalized image of the gel. A cubic spline is fitted through the pixel intensity values and a normalized image is extracted. This is even capable of converting an ABI gel into a conventional gel image.

* quick link - http://q.sanger.ac.uk/hrduoh7h