The Similarity Matrix feature in FluoroFinder calculates and displays spectral similarity data between different fluorophores on a panel. This information helps researchers understand how similar the emission spectra of different fluorophores are, which helps them avoid potential issues related to spectral overlap.

The similarity data is represented as an n x n grid, where n is the number of colors on the panel. Each cell in the grid represents the spectral similarity between two fluorophores. A score of 0 means they are not at all similar. A score of 1.0 is perfectly similar. In FluoroFinder’s Similarity Matrix, pairs of fluorophores with a high similarity score are highlighted in read as a warning that this combination of fluorophores may not be ideal.

In FluoroFinder's Panel Builder app, a similarity matrix will look something like this:

While in the Spectra Viewer it will appear as follows:

The cosine similarity formula is a common way to measure the similarity between two vectors, in this case, the emission spectra of two fluorophores. The cosine similarity measures the cosine of the angle between the two vectors. If the vectors are similar, the angle between them is small, and the cosine is close to 1. Conversely, if the vectors are dissimilar, the angle between them is large, and the cosine is close to 0 or negative.

The Similarity value between two fluorophores on a machine is the cosine similarity between those two fluorophores' emissions arrays on that machine.

The Matrix Score calculation is based on the condition number of the similarity matrix. The condition number measures how well-conditioned a problem is, with a low condition number representing a well-conditioned problem and a high condition number representing an ill-conditioned problem.

To compute the Matrix Score for a given similarity matrix, FluoroFinder follows these steps: