Gilar, M.; Fridrich, J.; Schure, M. R.; Jaworski, A. *Analytical Chemistry* **2012**, *84*, 8722–8732.

In this recent paper the metrics used to calculate orthogonality in LC x LC were evaluated using retention data for 194 different peptides. Correlation coefficients, mutual information, box-counting dimensionality, and surface fractional coverage were all applied to determine the orthogonality of the peptide separations. In addition, six simulated data sets were used to determine how applicable each metric was at determining LC x LC orthogonality in other types of highly correlated separations, i.e., bananagrams.

The statistical metrics corresponding to the surface coverage method were found to be the most useful and understandable for the sample sets typically seen in practice. The Gilar, convex hull, and dimensionality box counting surface coverage methods were found to be intuitively easy to understand, but the degree of orthogonality was determined to be highly dependent on the discretization (i.e., binning) of the separation space and susceptible to overestimates due to outliers.

The authors advocate the use of the surface coverage method for determining orthogonality because of its usefulness in calculating the achievable peak capacity of the separation (see equation 2). It is still unclear which of the surface coverage methods correlates best with the overall performance of real separations.

## Leave a Reply

You must be logged in to post a comment.