Feedback

type to search

Compute similarity between two distance matrices

Asked by [ Editor ]

I'm looking for a method to calculate the similarity between two distance matrices. I have two distance matrices that represents some X-data and another that represents Y-data and my goal is to see which X-data that is most similar to the Y-data. I could do something like a PLS with multi-Y, but I don't want to influence the X-data matrixes.

What I have is cross-reactivity data for a set of proteins and ligands. One of the X-distance matrixes consists of sequence homologies between the proteins and the Y-matrix will contain the distances derived from a PCA describing the ligand activities. I want to test another way of calculating the protein similarities and see if that measure is more correlated to the ligand activities. In the end, my goal is to take a new protein sequence, run it through the model and get suggestions on likely ligands.

Imported from: http://blueobelisk.stackexchange.com/questions/286

NN comments
chem-bla-ics
-

I guess the matrices are not of equal dimensions?

fredrik wallner
-

Yes, the distance matrixes are of equal dimensions.

or Cancel

1 answer

1

rajarshi guha [ Editor ]

You could try the Frobenious norm (http://en.wikipedia.org/wiki/Matrix_norm)

NN comments
fredrik wallner
-

If I have understood the concept of matrix norms correctly, it’s not exactly what I am after. As I understood it, the norm is a measure of the “area” of the matrix and doesn’t include information on which parts of the matrix contain high and low values. I tried to clarify my question a bit to explain what I am trying to do.

fredrik wallner
-

On second thought I think that I should have understood it as a calculation of the norm of the differences between the matrices. I will try that, than you for the suggestion.

or Cancel

Your answer

You need to join Blue Obelisk eXchange to complete this action, click here to do so.