Hi Anderson,
I am receiving a warning in PALM for one design matrix.
To provide you with some background, I am combining data from 2 studies. In each study, there are patients and controls, scanned at 2 time points (T1 & T2). Two sessions of resting state data were gathered at each visit for each participant.
In this analysis, I am only looking at data from subjects who completed both T1 and T2.
In my design matrix, I am currently using the first two EVs to represent diagnosis. A ‘-1’ in EV1 represents a patient scan at T1, and a ‘1’ in EV1 representing a patient scan at T2. Similarly, a ‘-1’ in EV2 represents a control subject
scan at T1, and a ‘1’ in EV2 represents a control subject scan at T2. I also have a covariate to represent study (the last EV). I have based the rest of the matrix on conversations I have seen at this forum. I am including an example matrix of 5 subs below,
representing 2 subs from study 1 -- 1 patient and 1 control – and 3 subs from study 2 – 2 patients and 1 control. I am also including an example of how the contrasts were specified below the matrix.
I am running PALM, using sign flippings and specifying the variance groups. I am getting the following warnings when running this analysis:
Warning: Matrix is close to singular or badly scaled. Results may be inaccurate. RCOND = 7.937625e-18.
3.68% [Shuffling 368/10000, Design 1/1, Contrast 3/4, Modality 1/21]
> In palm_core (line 1417)
In palm (line 81)
Warning: Rank deficient, rank = 79, tol = 8.652809e-13.
3.68% [Shuffling 368/10000, Design 1/1, Contrast 3/4, Modality 2/21]
> In palm_core (line 1417)
In palm (line 81)
Any feedback is much appreciated!!!
Best,
Tamara
Matrix:
|
Sub ID K01 -1 |
0 |
1 |
0 |
0 |
0 |
0 |
1 |
|
K01 -1 |
0 |
1 |
0 |
0 |
0 |
0 |
1 |
|
K01 1 |
0 |
-1 |
0 |
0 |
0 |
0 |
1 |
|
K01 1 |
0 |
-1 |
0 |
0 |
0 |
0 |
1 |
|
K02 0 |
-1 |
0 |
1 |
0 |
0 |
0 |
1 |
|
K02 0
|
-1 |
0 |
1 |
0 |
0 |
0 |
1 |
|
K02 0 |
1 |
0 |
-1 |
0 |
0 |
0 |
1 |
|
K02 0 |
1 |
0 |
-1 |
0 |
0 |
0 |
1 |
|
M01 -1 |
0 |
0 |
0 |
1 |
0 |
0 |
0 |
|
M01 -1 |
0 |
0 |
0 |
1 |
0 |
0 |
0 |
|
M01 1
|
0 |
0 |
0 |
-1 |
0 |
0 |
0 |
|
M01 1
|
0 |
0 |
0 |
-1 |
0 |
0 |
0 |
|
M02 -1 |
0 |
0 |
0 |
0 |
1 |
0 |
0 |
|
M02 -1 |
0 |
0 |
0 |
0 |
1 |
0 |
0 |
|
M02 1 |
0 |
0 |
0 |
0 |
-1 |
0 |
0 |
|
M02 1 |
0 |
0 |
0 |
0 |
-1 |
0 |
0 |
|
M03 0 |
-1 |
0 |
0 |
0 |
0 |
1 |
0 |
|
M03 0 |
-1 |
0 |
0 |
0 |
0 |
1 |
0 |
|
M03 0 |
1 |
0 |
0 |
0 |
0 |
-1 |
0 |
|
M03 0 |
1 |
0 |
0 |
0 |
0 |
-1 |
0 |
Contrasts:
|
T2>T1, G1 |
1 |
0 |
0 |
0 |
0 |
0 |
0 |
0 |
|
|
T2>T1, G2 |
0 |
1 |
0 |
0 |
0 |
0 |
0 |
0 |
|
|
Interaction (T2-T1, G1) > (T2-T1, G2) |
1 |
-1 |
0 |
0 |
0 |
0 |
0 |
0 |
|
|
Interaction (T2-T1, G1) > (T2-T1, G2) |
-1 |
1 |
0 |
0 |
0 |
0 |
0 |
0 |
|
From: "Sussman, Tamara (NYSPI)" <[log in to unmask]>
Date: Monday, March 19, 2018 at 9:46 AM
To: FSL - FMRIB's Software Library <[log in to unmask]>
Subject: Re: [FSL] Checking a design matrix for PALM
Thank you so much, Anderson!!
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: Sunday, March 18, 2018 at 5:21 PM
To: "[log in to unmask]" <[log in to unmask]>
Subject: Re: [FSL] Checking a design matrix for PALM
Test the intercept in PALM using sign-flippings (option -vg) and specify the variance groups with the option -vg. There is no need for exchangeability blocks then.