Confirming the PPI analysis steps

Dear AFNI experts,

I am looking for some advice regarding process the PPI analysis.
I followed the steps from “”.
However, I’m not sure that I conducted right, and there was several questions, which I didn’t understand.
(Even I checked the “-help” part, and so on.)

[Pre-processing information(steps) of my data]

  • despike, retroicor, tshift, align, tlrc, volreg, blur, mask, detrend, scale


  1. Generated a mask which I am interested in (i.e. ROI)

  2. Extract BOLD corresponding to the ROI
    3dmaskave -mask ACC+tlrc -quiet pb07.s10.r01.scale+tlrc > Seed_ACC.1D

  3. Remove the trend from the seed time series
    3dDetrend -polort 3 -prefix Seed_ACC_det Seed_ACC.1D'

  4. Transpose
    1dtranspose Seed_ACC_det.1D Seed_ACC_ts.1D

  5. Generate hemodynamic function
    waver -dt 2 -GAM -inline 1@1 > HRF.1D

    Q. When I conduct the GLM, I used WAV(15). Thus, in this case, which one is better to use, WAV or GAM?

  6. Conduct deconvolution of the seed time series
    3dTfitter -RHS Seed_ACC_ts.1D -FATUNG HRF.1D SeedNeur_ACC 012 0 - what is RHS, 012, 0??

    Q. As I understood, in option “-FALTUNG”, we need to input such as “012, 0” which is related to penalty function.
    However, even I checked the -help of 3dTfitter, I’m still not clear about the penalty function… Thus, may I ask you what is the penalty function in this

  7. Measure interaction regressor
    1deval -a SeedNeur_ACC.1D' -b task.1D -expr ‘a*b’ > Inter_ACC.1D

  8. Create interaction
    waver -GAM -peak 1 -TR 2 -input Inter_ACC.1D -numout 160TRs > Inter_ACC_2.1D

  9. Use Inter_A as an regressor for GLM analysis
    3dDeconvolve -input pb07.s10.r01.scale+tlrc.HEDA
    -censor motion_s10_censor.1D
    -polort 0
    -num_stimts 34
    -stim_times 1 stimuli/blk1.txt ‘WAV(15)’
    -stim_label 1 blk1
    -stim_times 2 stimuli/blk2.txt ‘WAV(15)’
    -stim_label 2 blk2

    -stim_label 20 blk20
    -stim_file 21 motion_demean.1D’[0]’ -stim_base 21 -stim_label 21 roll_01
    -stim_file 22 motion_demean.1D’[1]’ -stim_base 22 -stim_label 22 pitch_01
    -stim_file 23 motion_demean.1D’[2]’ -stim_base 23 -stim_label 23 yaw_01
    -stim_file 24 motion_demean.1D’[3]’ -stim_base 24 -stim_label 24 dS_01
    -stim_file 25 motion_demean.1D’[4]’ -stim_base 25 -stim_label 25 dL_01
    -stim_file 26 motion_demean.1D’[5]’ -stim_base 26 -stim_label 26 dP_01
    -stim_file 27 motion_deriv.1D’[0]’ -stim_base 27 -stim_label 27 roll_02
    -stim_file 28 motion_deriv.1D’[1]’ -stim_base 28 -stim_label 28 pitch_02
    -stim_file 29 motion_deriv.1D’[2]’ -stim_base 29 -stim_label 29 yaw_02
    -stim_file 30 motion_deriv.1D’[3]’ -stim_base 30 -stim_label 30 dS_02
    -stim_file 31 motion_deriv.1D’[4]’ -stim_base 31 -stim_label 31 dL_02
    -stim_file 32 motion_deriv.1D’[5]’ -stim_base 32 -stim_label 32 dP_02
    -stim_file 33 Seed_ts.1D -stim_label 33 Seed
    -stim_file 34 Inter_ACC_2.1D -stim_label 34 PPI_ACC
    -rout -tout
    -bucket PPI_stats.s10

  10. Group analysis

    Q. Am I correct to use “PPI_ACC#0_Coef” (the result of GLM in PPI) for t-test or ANOVA for a group analysis?

Thank you in advance.


DaWoon, your analysis looks fine to me. For future reference, you many follow the suggestion in the following thread:,154261,154416

Dear Gang,

Thank you for the confirmation and additional information related to the PPI analysis.
It’s really helped me to understand more. Thank you :slight_smile: