 |
 |
Dear Tom, Thanks for sending. For the 1st 3 contrasts, don't use "--permuteBlocks", as in this case you'd like to permute observations within block (within subject). I think I commented this in an earlier email. The simplest way to accomplish this is splitting the contrast file and running randomise twice, once with the "--permuteBlocks", another without. All the best, Anderson On 8 January 2015 at 10:06, Thomas Nickson <[log in to unmask]> wrote: > Hi Anderson, > > I have 266 scans in total, there are 48 controls, 68 high risk and 22 with > MDD. > > I'll also paste the con at the bottom and attach the other files. > > The command I'm using is: > > randomise -i ./stats/GM_mod_merg_s3.nii.gz -m > ./masks/AnteriorCingulateGyrusMask.nii.gz -o > ./AnteriorCingulateGyrus/T/fslvbm -d > ./design/RepeatedMeasure/finalDesign.mat -t > ./design/RepeatedMeasure/design.con -e ./design/RepeatedMeasure/design.grp > -c 2.3 -n 1000 --permuteBlocks > > Thank you for your time, > > Tom > > > ================================================================================================== > /ContrastName1 Time2 > Time1 G1 > /ContrastName2 Time2 > Time1 > G2 > /ContrastName3 Time2 > Time1 G3 > /ContrastName4 G2>G1 > /ContrastName5 G3>G1 > /ContrastName6 G1>G2 > /ContrastName7 G3>G2 > /ContrastName8 G1>G3 > /ContrastName9 G2>G3 > > /NumWaves 136 > /NumContrasts 9 > /RequiredEffect 0.954 0.954 0.535 0.79 0.917 0.917 0.71 0.71 0.467 > > /Matrix > 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > -1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > -1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 1 -1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 -1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 1 0 -1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 1 -1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 > > > > On Thu, Jan 8, 2015 at 9:01 AM, Anderson M. Winkler < > [log in to unmask]> wrote: > >> Dear Tom, >> >> Could you paste here the exact contents of the design.con file, and also >> the command line you are using to call randomise? >> Also, how many subjects do you have in each group? >> >> Thanks! >> >> All the best, >> >> Anderson >> >> >> On 7 January 2015 at 12:43, Thomas Nickson <[log in to unmask]> >> wrote: >> >>> Hello, >>> >>> I ended up having to permute the rows so that the timepoints were next >>> to each other i.e >>> >>> Grp G1 G2 G3 s1 s2 s3 s4 s5 s6 >>> 1 -1 0 0 1 0 0 0 0 0 >>> 1 1 0 0 1 0 0 0 0 0 >>> 2 -1 0 0 0 1 0 0 0 0 >>> 2 1 0 0 0 1 0 0 0 0 >>> 3 0 -1 0 0 0 1 0 0 0 >>> 3 0 1 0 0 0 1 0 0 0 >>> 4 0 -1 0 0 0 0 1 0 0 >>> 4 0 1 0 0 0 0 1 0 0 >>> 5 0 0 -1 0 0 0 0 1 0 >>> 5 0 0 1 0 0 0 0 1 0 >>> 6 0 0 -1 0 0 0 0 0 1 >>> 6 0 0 1 0 0 0 0 0 1 >>> >>> Does this remain correct? >>> >>> I keep everything else the same. >>> >>> I think this could be wrong because when I run it randomise with 100 >>> permutations tells me that: >>> >>> 100 permutations required for exhaustive test of t-test 1 >>> Doing all 100 unique permutations >>> >>> Are the groupings still correct? I don't know how it estimates the >>> number of permutations. >>> >>> Thanks, >>> >>> Tom >>> >>> >>> >>> >>> >>> >>> On Thu, Dec 11, 2014 at 3:20 PM, Anderson M. Winkler < >>> [log in to unmask]> wrote: >>> >>>> Hi Tom, >>>> It's probably better to save the computational time and run randomise >>>> twice, first with just within-block permutation, then add the other >>>> contrasts and run again with --permuteBlocks and --skipto=4, so that only >>>> the contrasts C4 onwards are performed. This will name the files >>>> consistently and prevent overwriting. >>>> All the best, >>>> Anderson >>>> >>>> >>>> On 11 December 2014 at 09:53, Thomas Nickson <[log in to unmask]> >>>> wrote: >>>> >>>>> Thanks for the reply Anderson! So should I split up the contrast file >>>>> so that I have one with the first three and another with remaining >>>>> contrasts or just run them all together with the two options? >>>>> >>>>> Thanks, >>>>> >>>>> Tom >>>>> >>>>> On Wed, Dec 10, 2014 at 6:16 PM, Anderson M. Winkler < >>>>> [log in to unmask]> wrote: >>>>> >>>>>> Hi Tom, >>>>>> >>>>>> Design and contrasts are fine. For the first 3 contrasts, you can use >>>>>> "-e design.grp" so that permutations will happen only within block. For the >>>>>> remaining contrasts, use also "--permuteBlocks". >>>>>> >>>>>> The groups can be unbalanced, but the number of timepoints need to be >>>>>> the same for all subjects. >>>>>> >>>>>> All the best, >>>>>> >>>>>> Anderson >>>>>> >>>>>> On 9 December 2014 at 15:51, Thomas Nickson <[log in to unmask] >>>>>> > wrote: >>>>>> >>>>>>> Hi All, >>>>>>> >>>>>>> I was looking to do a 3 factor, 2 level repeated measure analysis in >>>>>>> randomise for some VBM that I am doing and thought that I would run the >>>>>>> design matrix past the list to make sure that it's okay. >>>>>>> >>>>>>> I looked through the list and found this >>>>>>> >>>>>>> >>>>>>> https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1005&L=FSL&D=0&1=FSL&9=A&J=on&K=2&X=190C12580BEC088223&Y=thomas.nickson%40gmail.com&d=No+Match%3BMatch%3BMatches&z=4&P=343446 >>>>>>> >>>>>>> and thought that it would be possible to extend it in the following >>>>>>> way from 2 to 3 factors. >>>>>>> >>>>>>> For 6 subjects I get the following design matrix: >>>>>>> >>>>>>> Grp G1 G2 G3 s1 s2 s3 s4 s5 s6 >>>>>>> 1 -1 0 0 1 0 0 0 0 0 >>>>>>> 2 -1 0 0 0 1 0 0 0 0 >>>>>>> 3 0 -1 0 0 0 1 0 0 0 >>>>>>> 4 0 -1 0 0 0 0 1 0 0 >>>>>>> 5 0 0 -1 0 0 0 0 1 0 >>>>>>> 6 0 0 -1 0 0 0 0 0 1 >>>>>>> 1 1 0 0 1 0 0 0 0 0 >>>>>>> 2 1 0 0 0 1 0 0 0 0 >>>>>>> 3 0 1 0 0 0 1 0 0 0 >>>>>>> 4 0 1 0 0 0 0 1 0 0 >>>>>>> 5 0 0 1 0 0 0 0 1 0 >>>>>>> 6 0 0 1 0 0 0 0 0 1 >>>>>>> >>>>>>> Where, Grp is the group variable, G's are the groups and s is for >>>>>>> the subject variables. The negative numbers in the group column give T1 and >>>>>>> non-negative give T2. >>>>>>> >>>>>>> Contrasts: >>>>>>> >>>>>>> G1 G2 G3 s1 s2 s3 s4 s5 s6 >>>>>>> 1 0 0 0 0 0 0 0 0 Time2 > Time1 >>>>>>> G1 >>>>>>> 0 1 0 0 0 0 0 0 0 Time2 > Time1 >>>>>>> G2 >>>>>>> 0 0 1 0 0 0 0 0 0 Time2 > Time1 >>>>>>> G3 >>>>>>> -1 1 0 0 0 0 0 0 0 G2>G1 >>>>>>> -1 0 1 0 0 0 0 0 0 G3>G1 >>>>>>> 1 -1 0 0 0 0 0 0 0 G1>G2 >>>>>>> 0 -1 1 0 0 0 0 0 0 G3>G2 >>>>>>> 1 0 -1 0 0 0 0 0 0 G1>G3 >>>>>>> 0 1 -1 0 0 0 0 0 0 G2>G3 >>>>>>> >>>>>>> Is this correct? >>>>>>> >>>>>>> Also, is this only possible in a balanced design where there are >>>>>>> equal number of people at both time points? >>>>>>> >>>>>>> Thanks, >>>>>>> >>>>>>> Tom >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>> >>>>> >>>> >>> >> >