Error in CLR(context likelihood of relatedness)

Question

[nazaninhoseinkhan]

Hi all,

I am trying to work with CLR to obtain mutual information values. My input is the gene expression compendia of a bacterium. When I insert the command: [A, MIfull] = clr(data);, I receive this error:

Non-square or asymmetric input matrix; assuming unprocessed data.
Computing mutual information - this will take a while!
??? Undefined function or method 'mi' for input arguments of type 'double'.
Error in ==> clr at 64
MI = mi(data, n, k);

Can any one help me to find the reason of this error? Does it concern with my input?

Regards

Nazanin

Answer 1

[pegah]

Hi there,

The complete CLR code for running it in MATLAB for the expression data is something like the following code:

[A, MIfull] = clr(data, 'normal', 10, 3)
thresholdFDR = FDR(A, 0.05);
counter=1;
for i=1:4044
for j=i+1:4044
if(A(i,j)>=thresholdFDR)
result(counter,1)=i;
result(counter,2)=j;
counter=counter+1;
end
end
end

Where 4044 is an example for the number of row on the expression data, besides, 10 and 3 are number of bins and spline orders, respectively.

It seems to be funny but sometimes this error happen if you use CLR code in MATLAB in windows 64 bit. However, the algorithm run in win32 with no problem. It would be nice if someone tell me how I can revise the CLR code for win64.

Good luck,
Pegah