-1 1 1 1-1 1 2 1-1 1 3 1-1 1 4 1-1 1 5 1-1 2 1 1-1 2 1 2-1 2 2 1-1 2 2 2-1 3 1 1-1 3 1 2-1 3 1 3-1 3 2 1-1 3 2 2-1 3 2 3-1 3 3 1-1 3 3 2-1 3 3 3
Hi Anderson,Thanks for your reply. Could you please verify is the below EB is correct, for two families of two siblings, three families of three siblings and five independent individuals? I expect this EB design to permute whole blocks of similar families, and within blocks of all families and independent individuals.All the best,Sourena-1 1 1-1 1 1-1 1 1-1 1 1-1 1 1-1 2 1-1 2 1-1 2 2-1 2 2-1 3 1-1 3 1-1 3 1-1 3 2-1 3 2-1 3 2-1 3 3-1 3 3-1 3 3Hi Sourena,You can, but then you need to use multi-level exchangeability blocks (as describe in the HCP permutation paper: https://doi.org/10.1016/j.neuroimage.2015.05.092It can't be done with simple EBs. In Figure 1, that is the "notation using a single column".All the best,AndersonHi Anderson,Thanks for your helpful clarification. I have one follow up question: Considering that I have many exchangeability blocks of different sizes (i.e. family structures), but many of these EBs are of the same size, is it not possible to do both within block and whole-block permutations among those families with similar sizes? I get the below error when I try both -within and -whole arguments:Not all sub-blocks within an EB are of the same size at level 1.All the best,SourenaHi Sourena,Exactly, there is no need in principle to correct for different variances in this case.The -vg is for different variances within subject. For example, suppose that you are studying weight and some subjects were measured using a low quality scale that had a lot of variability just due to the instrument's internal mechanical or electronic parts, whereas some other subjects were measured using a more robust scale that is not quite as prone to such variation. Another case could be if some subjects were measured just once, whereas others were measured multiple times and those values then averaged within person. Then those measured just once would be expected to have a higher variance than those who had their measurements averaged after multiple trials.Lower variability due to family relationships have to do with heritability and isn't really a case to use variance groups.All the best,AndersonHi,In a rather large patient/control cohort, in which many (but not all) of the subjects are siblings of different family sizes (mostly 2 siblings), I wonder if I should control for heteroscedasticity by separating variance groups in PALM, since the brain phenotype variance is expected to be lower between subjects of the same genetic background.But reading the FSL PALM webpage, it is advised against this for the HCP data:
"unless a custom file is supplied to PALM with -vg, or if -vg auto is used, in which case, the most restrictive possible variance group configuration will automatically be created, using the information from the exchangeability blocks file. Typically you won't want to do this with HCP data, so that the option -vg can be omitted".Can someone enlighten me what is the rationale behind this approach?
Cheers,Sourena
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1