I'm working with the algorithm presented in the paper Inference of gene regulatory networks and compound mode of action from time course gene expression profiles. This is an implementation of TSNI in MATLAB which takes as input a n x m gene expression matrix (n is the number of genes and m the number of time measurements in the experiment) and outputs an n x n weighted adjacency matrix.
In the supplementary materials webpage, the authors state that the first set of time points should be scaled to zero (the first column of the input matrix) and presumably this also means the rest of the columns go through some sort of transformation as well depending on which set of time points they represent (1, 2, 3, 4, etc.) although I'm not certain about this. My question is precisely how to carry out this scaling process and what computation exactly is to be made for this as I couldn't find any information regarding this anywhere (even the paper doesn't go over this).
The authors provide a demo along with the code and the first column of the input matrix is indeed all zeros, but I have no way to know if the rest of the columns were modified or how. I'm by no means an expert on TSNI or similar processes and I can't tell if it makes sense to just set the first column to zeros and leave the rest as it is or if each column has to go through some scaling process. I leave a capture of the demo input matrix for reference (9 genes across 6 different sets of time points).
edit: it also occurred to me at some point that maybe the column of zeros is just an extra artificial column that gets added at the beginning of the matrix, but again I don't know for sure.
wouldn't this be dividing by zero though?