3dLME script

AFNI Message Board
1

Hi AFNI,

I’m trying to run a linear mixed model using 3dLME between 2 groups across 3 different sessions/time points (some subjects have missing data) while controlling for age, sex, and framewise displacement. I just want to make sure my script looks ok. I’m trying to model a group x session interaction while controlling for the aforementioned variables. With ~1+RT, Im specifying different intercepts and slopes for subjects. Is there something I’m missing?

module load rc/afni/20.1.18-foss-2017b-Python-2.7.14
module load R/3.2.0-goolf-1.7.20

3dLME -prefix /FC_data/group_analysis/results_3dLME_DMN.nii -jobs 10
-model ‘Group*Session+Age+Sex+FD’
-qVars ‘Age,FD’
-ranEff ‘~1+RT’
-SS_type 3
-dataTable
Subj Group Session Age Sex FD InputFile
sub-control01 HC 1 28 2 0.1217738 /FC_data/DMN/connmap_sub-control01_ses-01_DMN.nii
sub-control01 HC 2 28 2 0.1749546 /FC_data/DMN/connmap_sub-control01_ses-02_DMN.nii
sub-control01 HC 3 28 2 0.150672 /FC_data/DMN/connmap_sub-control01_ses-03_DMN.nii
sub-control02 HC 1 23 1 0.1281235 /FC_data/DMN/connmap_sub-control02_ses-01_DMN.nii
sub-control03 HC 1 25 1 0.1455679 /FC_data/DMN/connmap_sub-control03_ses-01_DMN.nii
sub-control03 HC 2 25 1 0.3276477 /FC_data/DMN/connmap_sub-control03_ses-02_DMN.nii
…
sub-patient01 FEP 1 25 1 0.1971614 /FC_data/DMN/connmap_sub-patient01_ses-01_DMN.nii
sub-patient01 FEP 2 25 1 0.2590447 /FC_data/DMN/connmap_sub-patient01_ses-02_DMN.nii
sub-patient01 FEP 3 25 1 0.254582 /FC_data/DMN/connmap_sub-patient01_ses-03_DMN.nii
sub-patient02 FEP 2 30 2 0.1924839 /FC_data/DMN/connmap_sub-patient02_ses-02_DMN.nii
…

Thanks,
O.

2

I just want to make sure my script looks ok.

Change

-ranEff '~1+RT' \

to

-ranEff '~1+FD' \

3

Hi again,

I want to compare each group at each time point. Is this how my glts should look like?

-num_glt 3
-gltLabel 1 ‘HC-FEP_ses-01’ -gltCode 1 ‘Session: 1 : 1HC -1FEP’
-gltLabel 2 ‘HC-FEP_ses-02’ -gltCode 2 ‘Session 2 : 1HC -1FEP’
-gltLabel 3 ‘HC-FEP_ses-03’ -gltCode 3 ‘Session 3 : 1HC -1FEP’ \

4

Since session is a factor. I suggest that you code session with something like S1, S2, and S3 (instead of 1, 2, and 3)so that it would avoid some confusion. Specify a comparison with something like

-gltCode 1 ‘Session: 1S1 Group : 1HC -1*FEP’

5

-gltCode 1 ‘Session: 1S1 Group : 1HC -1FEP’
-gltCode 2 'Session: 2S2 Group : 1HC -1FEP’
-gltCode 3 ‘Session: 3S3 Group : 1HC -1*FEP’

So, this is how it would look for all 3 comparisons then?

Thanks a lot!

6

Leave an empty space after ‘Session’ and keep the weight to 1 for each session:

-gltCode 1 ‘Session : 1S1 Group : 1HC -1FEP’
-gltCode 2 'Session : 1S2 Group : 1HC -1FEP’
-gltCode 3 ‘Session : 1S3 Group : 1HC -1*FEP’