Definitions.

The following are brief definitions of the concepts of spot and pair matching correctness.

Spot correctness

Spot correctness can be defined as the ratio of correct spots (according to visual criteria) to the total number of spots evaluated. Below follow the visual criteria used.

A correct spot.

correct_spot.png

A correct spot is defined as a single, real protein spot with a border that is encompassing the entire spot, but no part of another spot or artifacts in the image. See image example.

However, a border is allowed to contain background surrounding the spot, since all software packages are able to subtract background intensity from the cumulative spot volume.

An incorrect spot.

Basically, there are three types of incorrect spots: false, misshaped and missing.

a) A false spot is a spot border around something that obviously is not protein material, such as an artefact in the gel or similar.

false_spot.png

b) A misshaped spot is a spot that has some sort of error in the spot border. Either it does not encompass the entire spot body properly, or it encompasses surrounding spots, or a combination thereof. Both cases give rise to a faulty measurement of the spot intensity.

misshaped_spot.png

c) A missing spot is a spot that has not been detected by the software.

missing_spot.png

Using these definitions we then define Spot Correctness as follows:

 

spot correctness = correct spots / spots evaluated


Pair matching correctness

Match correctness can be defined as the ratio of correct pair matchings (according to visual criteria) to the total number of matchings.

The Combined Correctness evaluation aims to discriminate between matching performance and spot detection performance as much as possible. For this reason, only cases where the detection of the spots in both gels was correct should be considered, where the program had at least a chance to match correctly. In this way, non-decidable or borderline cases can to a large extent be avoided.

A correct pair matching.

a) A pair matching is said to be correct when a spot in the first gel image has been matched to a spot in the second gel image that according to visual inspection appears to be the corresponding spot. (correct matching)

correct_matching.png

b) It is also said to be correct when there is NO matching and NO corresponding spot in the other image. (correct non-matching)

correct_non_matching.png

An incorrect pair matching.

a) A pair matching is said to be incorrect when a spot in the first gel image has been matched to a spot in the second gel image that does NOT appear to be the same spot. (incorrect matching, or “mis-match”)

incorrect_matching.png

b) It is also said to be incorrect if there is NO matching but there DOES appear to be a corresponding spot in the second image (incorrect non-matching, or “missing matching”).

incorrect_non_matching.png

Using these definitions, we can then define the pair matching correctness as:

 

pair matching correctness =  correct matchings / matchings evaluated

A few comments

This method of evaluating pair matching correctness attempts to consider matching at the most basic level: between a pair of spots. The results will consequently be comparable between projects with different numbers of gels.

If instead we had opted to look at the correctness of matching of a protein through all the gels in a project, then clearly the result would mean something very different in a project with a large number of gels compared to a smaller project, because the risk of a matching error increases with the number of gels. 

 

Download the free Combined Correctness Tool for 2D gel data QC