Hi Heidi, the problem is being caused by the interaction column. Try demeaning the ed_log and MMSE before multiplying them together to get the interaction. Do the demeaning on the entire data set, not on a group by group basis. doug
On 08/27/2013 02:04 AM, Jacobs H (NP) wrote:
Sure, here it is. Thanks Heidi ================================================ Dr. Heidi Jacobs Postdoc researcher Faculty of Health, Medicine and Life Sciences School for Mental Health and Neurosciences Division Cognitive Neuropsychiatry and Clinical Neurosciences Alzheimer Center Limburg h.jacobs@maastrichtuniversity.nl www.maastrichtuniversity.nl www.heidijacobs.nl
Dr. Tanslaan 12, 6229 ET Maastricht P.O. Box 616, 6200 MD Maastricht, The Netherlands T +31 43 38 84 090 F +31 43 38 84 092 ================================================ ________________________________________ Van: Douglas N Greve [greve@nmr.mgh.harvard.edu] Verzonden: maandag 26 augustus 2013 23:53 Aan: Jacobs H (NP) CC: freesurfer@nmr.mgh.harvard.edu Onderwerp: Re: matrix ill-conditioned....
Can you send me the design matrix (Xg.dat) and fsgd file?
On 08/26/2013 04:35 PM, Jacobs H (NP) wrote:
Dear Doug,
Thank you. I tried to run it with this version of mri_glmfit (saved as mri_glmfit2), but it still gives me the same error message. I don't see the problem. I already divided the column with the interaction by 10. Fsgd files without MMSE and interaction factors, did not gave a problem... Could you have a look at it? I will attach the fsgd and contrast file.
Thanks! I really appreciate your help! Heidi
ERROR: matrix is ill-conditioned or badly scaled, condno = 101281
Possible problem with experimental design: Check for duplicate entries and/or lack of range of continuous variables within a class. If you seek help with this problem, make sure to send: 1. Your command line: mri_glmfit2 --y rh.Alls_educlog_MMSE_inter_agecent_ICVcent_fsgd.sm20.mgh --fsgd /media/Data_disk/Nieuwe_map/FreeSurferOutputAD/FreeSurferOutputv450/FSGD/fsgd.Alls_educlog_MMSE_inter_agecent_ICVcent_fsgd.txt dods --glmdir rh.Alls_educlog_MMSE_inter_agecent_ICVcent_fsgd.dods.20.glmdir --surf fsaverage rh --C /media/Data_disk/Nieuwe_map/FreeSurferOutputAD/FreeSurferOutputv450/GLM_CONTRAST.MAT/contrast_3groups_Finteraction.mat 2. The FSGD file (if using one) 3. And the design matrix above
================================================ Dr. Heidi Jacobs Postdoc researcher Faculty of Health, Medicine and Life Sciences School for Mental Health and Neurosciences Division Cognitive Neuropsychiatry and Clinical Neurosciences Alzheimer Center Limburg h.jacobs@maastrichtuniversity.nl www.maastrichtuniversity.nl www.heidijacobs.nl
Dr. Tanslaan 12, 6229 ET Maastricht P.O. Box 616, 6200 MD Maastricht, The Netherlands T +31 43 38 84 090 F +31 43 38 84 092 ================================================ ________________________________________ Van: Douglas N Greve [greve@nmr.mgh.harvard.edu] Verzonden: maandag 26 augustus 2013 21:47 Aan: Jacobs H (NP) CC: freesurfer@nmr.mgh.harvard.edu Onderwerp: Re: matrix ill-conditioned....
You can try with this version of mri_glmfit which will do the rescaling for you ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/mri_glmfit
If the problem persists then it may be an issue with your design
doug
On 08/26/2013 01:42 PM, Jacobs H (NP) wrote:
Hi Doug,
I divided everything by 10, but the problem remains. I have been searching a bit more. Could it be that the problem is that the range in cognitive scores in the control group is very small? When I run the groups separately, the patients are not a problem (the range in cognition is very large), but the controls are still the problem.
Thanks! Heidi ================================================ Dr. Heidi Jacobs Postdoc researcher Faculty of Health, Medicine and Life Sciences School for Mental Health and Neurosciences Division Cognitive Neuropsychiatry and Clinical Neurosciences Alzheimer Center Limburg h.jacobs@maastrichtuniversity.nl www.maastrichtuniversity.nl www.heidijacobs.nl
Dr. Tanslaan 12, 6229 ET Maastricht P.O. Box 616, 6200 MD Maastricht, The Netherlands T +31 43 38 84 090 F +31 43 38 84 092 ================================================ ________________________________________ Van: Douglas N Greve [greve@nmr.mgh.harvard.edu] Verzonden: vrijdag 23 augustus 2013 17:31 Aan: Jacobs H (NP) CC: freesurfer@nmr.mgh.harvard.edu Onderwerp: Re: [Freesurfer] FW: FW: Contrast and F-test
The values in the interaction column are probably much larger than the values in the other columns (thus they are "badly scaled"). You can divide them by a number to bring their scale down to that of the other columns without changing the p-values. Version 5.x does this automatically.
On 08/23/2013 11:25 AM, Jacobs H (NP) wrote:
I am using version 4.5, as all the old analyses were also done with this version. What do you mean with rescale? THanks Heidi ________________________________________ Van: freesurfer-bounces@nmr.mgh.harvard.edu [freesurfer-bounces@nmr.mgh.harvard.edu] namens Douglas N Greve [greve@nmr.mgh.harvard.edu] Verzonden: vrijdag 23 augustus 2013 17:23 Aan: freesurfer@nmr.mgh.harvard.edu Onderwerp: Re: [Freesurfer] FW: FW: Contrast and F-test
what version of FS are you using? It may be that you just need to rescale that column (if the scale is much more than the other columns)
On 08/23/2013 11:12 AM, Jacobs H (NP) wrote:
Hi again!
I am trying to run these contrasts but FreeSurfer tells me that "ERROR: matrix is ill-conditioned or badly scaled, condno = 101380 Possible problem with experimental design: Check for duplicate entries and/or lack of range of continuous variables within a class."
Could this be due to the fact that there is a high correlation between cognition, education and cognition x education? (the product is correlated with the factors of that product) Or is something else going on? How can I solve this?
Thanks Heidi
Van: freesurfer-bounces@nmr.mgh.harvard.edu [freesurfer-bounces@nmr.mgh.harvard.edu] namens Douglas N Greve [greve@nmr.mgh.harvard.edu] Verzonden: donderdag 22 augustus 2013 17:52 Aan: freesurfer@nmr.mgh.harvard.edu Onderwerp: Re: [Freesurfer] FW: FW: Contrast and F-test
both of those contrasts are correct doug
On 08/22/2013 10:51 AM, Jacobs H (NP) wrote:
Arrghh, again wrong email, sorry....
Thanks. I think I understand the rational behind it. Just to be sure, would this be correct:
For each group separately 0 0 0 1 0 0 (group - cognition - education - cognition x education - age - ICV)
For the F-test (same order) 0 0 0 0 0 0 0 0 0 1 -1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 -1 0 0 0 0 0 0
Heidi
Van: freesurfer-bounces@nmr.mgh.harvard.edu [freesurfer-bounces@nmr.mgh.harvard.edu] namens Douglas N Greve [greve@nmr.mgh.harvard.edu] Verzonden: donderdag 22 augustus 2013 16:46 Aan: freesurfer@nmr.mgh.harvard.edu Onderwerp: Re: [Freesurfer] FW: Contrast and F-test
you need to have all 3. If you want to examine the interaction, then you need a new contrast to test the interaction. Set all to 0 except for the interaction term doug
On 08/22/2013 10:42 AM, Jacobs H (NP) wrote: > Dear Doug, > Thank you for the information. Sorry for not sending this to the list. > Do you mean that I then have to include the three variables (covariate 1, covariate 2, product of these two) or only the product in the fsgd file? > If I need to include all three of them, can you advise me on the contrasts? > If it is only the product, I think I can use the same contrasts as before, right? > > Thanks > Heidi > ================================================ > Dr. Heidi Jacobs > Postdoc researcher > Faculty of Health, Medicine and Life Sciences > School for Mental Health and Neurosciences > Division Cognitive Neuropsychiatry and Clinical Neurosciences > Alzheimer Center Limburg > h.jacobs@maastrichtuniversity.nl > www.maastrichtuniversity.nl > www.heidijacobs.nl > > Dr. Tanslaan 12, 6229 ET Maastricht > P.O. Box 616, 6200 MD Maastricht, The Netherlands > T +31 43 38 84 090 F +31 43 38 84 092 > ================================================ > ________________________________________ > Van: freesurfer-bounces@nmr.mgh.harvard.edu [freesurfer-bounces@nmr.mgh.harvard.edu] namens Douglas N Greve [greve@nmr.mgh.harvard.edu] > Verzonden: donderdag 22 augustus 2013 16:34 > Aan: freesurfer@nmr.mgh.harvard.edu > Onderwerp: Re: [Freesurfer] Contrast and F-test > > If you want to test for an interaction between continuous variables, > then you should create a new regressor by multiplying them together then > test that regressor > doug > > > On 08/22/2013 08:09 AM, Jacobs H (NP) wrote: >> Hi all, >> >> I my manuscript, I examined the association between cognition and cortical thickness. I first did this in three groups separately and then ran a F-test. >> One of the reviewers, now wants me to examine the interaction between cognition and education (all continuous). >> >> Would this be the correct contrast for each group separately (last two zeros are for covariates, also continuous) >> >> 0 1 -1 0 0 >> >> Would this be the correct way for the F-test: >> >> 0 0 0 0 0 1 -1 0 -1 1 0 0 0 0 0 0 0 >> 0 0 0 1 0 -1 -1 0 1 0 0 0 0 0 0 >> >> Thanks! >> >> Heidi >> >> ================================================ >> Dr. Heidi Jacobs >> Postdoc researcher >> Faculty of Health, Medicine and Life Sciences >> School for Mental Health and Neurosciences >> Division Cognitive Neuropsychiatry and Clinical Neurosciences >> Alzheimer Center Limburg >> h.jacobs@maastrichtuniversity.nl >> www.maastrichtuniversity.nl >> www.heidijacobs.nl >> >> Dr. Tanslaan 12, 6229 ET Maastricht >> P.O. Box 616, 6200 MD Maastricht, The Netherlands >> T +31 43 38 84 090 F +31 43 38 84 092 >> ================================================ >> _______________________________________________ >> Freesurfer mailing list >> Freesurfer@nmr.mgh.harvard.edu >> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer >> >> > -- > Douglas N. Greve, Ph.D. > MGH-NMR Center > greve@nmr.mgh.harvard.edu > Phone Number: 617-724-2358 > Fax: 617-726-7422 > > Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting > FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 > www.nmr.mgh.harvard.edu/facility/filedrop/index.html > Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/ > > _______________________________________________ > Freesurfer mailing list > Freesurfer@nmr.mgh.harvard.edu > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > > > The information in this e-mail is intended only for the person to whom it is > addressed. If you believe this e-mail was sent to you in error and the e-mail > contains patient information, please contact the Partners Compliance HelpLine at > http://www.partners.org/complianceline . If the e-mail was sent to you in error > but does not contain patient information, please contact the sender and properly > dispose of the e-mail. > > > _______________________________________________ > Freesurfer mailing list > Freesurfer@nmr.mgh.harvard.edu > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer >
>
Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/