Hi Anderson,

 

OK, I think I get this. Since I have a single modality, there is no need to include modality-related flags (don’t use -npcmod or -corrmod). And the overall effect of me including six designs ( six -d flags) and one t-contrast (one -t flag, containing “+1 /n -1”) results in a pooled net total of 12 contrasts. So I just need to include flags for testing/correcting/pooling over contrasts. The -npccon flag will do the joint inference over the 12 contrasts for my global null hypothesis test. And the -corrcon flag will do the corrections for multiple testing for each of my partial tests.

 

Pretty sure I’ve got that correct, but just let me know if that’s still off. Thank you very much, Anderson – this has been super helpful.

 

All the best,

Luke.

 

From: FSL - FMRIB's Software Library <[log in to unmask]> on behalf of "Anderson M. Winkler" <[log in to unmask]>
Reply to: FSL - FMRIB's Software Library <[log in to unmask]>
Date: Friday, 16 August 2019 at 13:42
To: "[log in to unmask]" <[log in to unmask]>
Subject: Re: [FSL] PALM NPC query - 1 modality, 2 contrasts, and 6 designs

 

Hi Luke,

 

I'm sorry -- in the first post, it's -npccon that you need, not -npcmod. The -npccon will do the joint inference over the 12 contrasts, therefore pooling together all the 6 designs.

 

There isn't a -corrdes or -npcdes option... combination and correction happen at the level of the input data, or of the hypotheses tested on that data; the latter are specified by the contrasts. The combination of designs is the combination of contrasts, in that it's the former that indicate how the latter are meant to be tested.

 

So it seems to me that all that you need is replace -corrmod for -corrcon and you should be ready to go.

 

All the best,

 

Anderson

 

 

On Thu, 15 Aug 2019 at 13:12, Luke Baxter <[log in to unmask]> wrote:

In short(er):

There is an "-npcmod" flag to enable NPC over modalities.
There is an "-npccon" flag to enable NPC over contrasts.
Is there an equivalent of an "-npcdes" flag to enable NPC over designs?

Since I have multiple contrasts and multiple designs, ideally I would use something like "-npccon" and "-npcdes". I presume "-npcmod" would do nothing in my case, since I only have a single modality. That is, I only have one -i input, and multiple -d inputs, and my -t input csv file has two rows (two contrasts).

The wiki says/implies I can do NPC over designs, but I can't see the appropriate flag. And as I mentioned in the last post, I can't see the relevant outputs from PALM.

To labor the point, correcting over modalities produces an mfwep file, correcting over contrasts produces a cfwep file, and correcting over both gives an mcfwep file. If correcting over designs (using the hypothetical -npcdes flag) produced a dfwep file, I believe I'm looking for something like a dcfwep file. For each partial test.

I'd also like the NPC global null (across contrasts and designs) equivalent to combining all designs into one and doing a global f-test.

Many thanks,
Luke.

########################################################################

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