I’m currently running a relatively complex model in nipype==1.4.2 that’s using the FSL specific SpecifyModel, Level1Design and FEATModel interfaces. The design matrix fed in to SpecifyModel interface contains 4 regressors for the first 4 legendre polynomials (0, 1, 2, 3), with the 0th being a one for every volume. When the final design matrix produced by FEATModel is produced the column associated with the 0th legendre polynomial has been converted to a 0 for all volumes. Does anyone have an idea of why this may be occurring?
A second related issue I’ve found while troubleshooting this issue is that the ‘.png’ representing the design matrix is producing output that suggests that the value in the matrix for that column is all ones, when they are in fact all 0 or close to zero.
For clarity, this is being run on a singularity image with Nipype 1.4.2 and FSL version 5.0.9 respectively. Attached are the design matrix and png in question! run0.txt (116.5 KB)