Permutations with Mask Roi

Written by Owen Phillips. Email Dr. Katherine Narr if you have any questions. For more information, see the protocols page.

Permutaions module running on the pipeline Permutations module executing.

Permutations with Mask Roi

  1. This protocol shows how to run permutations using roi masks.

    screenshot of ucf roi
  2. To run permutations you will need a the following things.

    A) A text file listing the *ucf's of interest with their corresponding variable values. To learn how to make this text file in the proper format. Read Creating a Text file .
    B) Your roi ucf masks. This protocol will use roi's from /cxfs/schizo/bilder/YALE_LABEL_ROIS. If you do not have access to these. Send me an email and I will try to help. Also you can make your own roi's.
    C) A working version of the LONI Pipeline. Developed by the great Pipeline team. (Alternatively you can run a command line based approach on cerebro. An example command is at the bottom of the page.)
    D) A Anova Permutations with Mask Pipeline Module. Available for download.
  3. Start the Pipeline program and open the Permutations with Mask module. The first input required is the text file.
    If your file is not on your local computer, click the remote option.
  4. Your next input is your formulay~. In this example I want the main effect of diagnosis covaried for age.
  5. The next input is your reducedy~. This will be whatever your covariates are.
  6. The mask input will be a *.ucf file. (or a *.list for advanced pipeline users)
  7. The threshold value is set at .5. The threshold value refers to the ROI masks. The masks are 4D hemisphere mesh ucfs. With the areas of interest having a value of 1 in the 4th column. In order for the permutations command to look within this area you have to give it a threshold value. It will look in the points where it is greater then .5. You can open up one of the ROI's in nedit and scroll down. You will see a grouping of points that have a value of 1. Everything else will have a value of 0 in the 4th column. So, if you gave it a threshold of 0 it would be as if you didn't give it a mask at all. Or if you gave it any value over 1 it would have the same effect.
  8. You can set the number of permutations you wish to perform. 1000 is standard in most cases.
  9. The pcutoff refers to the significance factor you are looking for in your given area. .05 and .01 are the most common.
  10. You should not need to change any of the other values. Click run. Takes about 5-15 mins depending on the number of permutations you are running.
  11. Your output will be a .txt file. To view your result scroll down to the bottom and it will have a line reading something like this:
    [1] "probability of reaching or exceeding observed number of points greater than cutoff:"
    [1] 0.13
    If you do not get the significant result you were hoping for, you can try changing the pcutoff and permutations number.


Sample permutations command line.


  1. qsub -b y -q long.q /cxfs/woods/R/R-2.4.1/bin/R CMD BATCH --no-save --no- restore --quiet --args -table/cxfs/schizo/bilder/ALL _HEM/ sod6grp_all_hem_L_categscantype.txt -formulay~sex+age +brainvol_cuberoot+ScanType+sod6grp -reducedy~sex+age +brainvol_cuberoot+ScanType -mask/cxfs/schizo/bilder/YALE_LABEL_ROIS/ avg_yale_176_lateral_dorsal_frontal_no_overlap.ucf -threshold0.5 - permutations10 -pcutoff0.05 /cxfs/woods/rshape/batch_commands/ anova_shape_masked_permute_batch.R /cxfs/schizo/bilder/ALL_HEM/ SOD6GRP_PMAPS/PERMUTATIONS/ lateral_dorsal_frontal_perms_mnsod_covar_sexagebvscantype_L_hem.txt