The Combined Correctness metric
Combined Correctness is the name of a metric for quality control of data from 2D gel image analysis. Combined Correctness is expressed as a percentage score, that allows us to draw statistically relevant conclusions about how good the 2D gel image analysis really is.
It can be interpreted as the likelihood that a randomly chosen true protein spot in one of the gels in the experiment is correctly detected and matched (to a corresponding true spot in another gel image) according to visual criteria.
Thus, in plain language it can be interpreted as the likelihood that the data coming from this measurement is reliable.
The Combined Correctness metric is the product of spot correctness and pair matching correctness (as described on the algorithms page) and is a metric that can be applied independent of the software used to generate the data.
In short:
| Total Spot Correctness = Correct / (Correct + False + Misshaped + Missing) |
| Pair matching correctness = Correct / (Correct + Incorrect) |
| Combined correctness = [Spot correctness] x [Pair matching correctness] |
Read more about the definitions of Spot Correctness and Match Correctness here.
Read more about the algorithm for how to assess Spot and Match Correctness here.
Higher Combined Correctness always means more true positives:
Above Figure:For 11 independent sets of 2D gel projects two image analyses were preformed using different software. A Combined Correctness score was calculated for each analysis and the number of true positives in the image analysis data estimated. In all cases, data sets with a higher Combined Correctness score also displayed a higher percentage of true positives.
Higher Combined Correctness always means a lower false positive rate:
Above Figure:When the Combined Correctness is improved, the false discovery rate goes down. The Pearsson correlation coefficient for Combined Correctness’ correlation to false positives rate was 0.47, indicating a strong relationship between the two.
All data taken from the poster How to Predict Discovery Potential and False Positives in 2D Electrophoresis Image Analysis presented at the 2007 HUPO.
Reference material for Combined Correctness
Combined Correctness Can Enrich Proteomics
Genetic Engineering News, April 2008
How to Predict Discovery Potential and False Positives in 2D Electrophoresis Image Analyses
HUPO 2007 Poster
Contact us for more information.
