 |
 |
Hi, It sounds like you have an extreme type of design where it cannot do more than 53 valid permutations. This can happen when, for example, one group only has one subject in it. In this case it is only possible to swap this one subject with each other subject in turn, and then that is the limit of all the possible permutations. The limit that you put in is a maximum number of requested permutations, but if the mathematics doesn't allow for more than a certain number then you cannot go beyond that. If you think that your design is not unusual then this might indicate an error in how you've set it up. Otherwise, you may be stuck with this many permutations and the consequent limitation in statistical power. All the best, Mark > On 20 Mar 2018, at 19:17, Paolo Nucifora <[log in to unmask]> wrote: > > I am confused by how randomise determines the number of permutations it will perform. > > Randomise is currently running as part of dual_regression over 53 fMRI scans. I specified a maximum of 5000 permutations, yet when I examined the logs I found that only 53 permutations were performed. > > Even stranger, running randomise from the command line with the same parameters produced the expected number of permutations for some (but not all) of the components. I examined the dr_stage2_ic files in fslview and there were no obvious issues. > > What is going on? > > > > $ cat drD.o22851.10 > randomise options: -i /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/dr_stage2_ic0009 -o /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/dr_stage3_ic0009 -m /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/mask -d /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/design.mat -t /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/design.con -n 5000 -T -V > Loading Data: > Data loaded > 53 permutations required for exhaustive test of t-test 1 > Doing all 53 unique permutations > Starting permutation 1 (Unpermuted data) > Starting permutation 2 > Starting permutation 3 > > > $ randomise -i /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/dr_stage2_ic0009 -o /home/nucifora/comp/melodic/redo/dualreg/test-dr_stage3_ic0019 -m /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/mask -d /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/design.mat -t /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/design.con -n 5000 -T -V > randomise options: -i /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/dr_stage2_ic0009 -o /home/nucifora/comp/melodic/redo/dualreg/test-dr_stage3_ic0019 -m /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/mask -d /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/design.mat -t /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/design.con -n 5000 -T -V > Loading Data: > Data loaded > 2756 permutations required for exhaustive test of t-test 1 > Doing all 2756 unique permutations > Starting permutation 1 (Unpermuted data) > Starting permutation 2 > Starting permutation 3 > > > $ randomise -i /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/dr_stage2_ic0019 -o /home/nucifora/comp/melodic/redo/dualreg/test-dr_stage3_ic0019 -m /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/mask -d /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/design.mat -t /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/design.con -n 5000 -T -V > randomise options: -i /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/dr_stage2_ic0019 -o /home/nucifora/comp/melodic/redo/dualreg/test-dr_stage3_ic0019 -m /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/mask -d /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/design.mat -t /home/nucifora/comp/melodic/redo/dualreg/n5000_20180320011258/design.con -n 5000 -T -V > Loading Data: > Data loaded > 53 permutations required for exhaustive test of t-test 1 > Doing all 53 unique permutations > Starting permutation 1 (Unpermuted data) > Starting permutation 2 > Starting permutation 3