correlation coefficients for a matrix that includes NaNs

I am trying to determine correlation coefficients and p values for different parameters of timeseries data. I have a matrix of five variables (each variable is a column and rows are times). I don't have data for every parameters at every timestamp so there are NaNs during the no value periods. Is there a function that will determine correlation coefficients for a dataset that includes NaNs? Right now I am using corrcoef and it gives me matrices for the pvalues and coefficients that are all 5 by 5 and filled with all NaNs. Any ideas?

Right now I am doing this: [R, P] = corrcoef(PK6mab) using the matrix in the mat file attached

Answer 1

[David Hill]

[R,P]=corrcoef(rmmissing(PK6mab));

If you can tolerate deleting the entire rows of any missed data.