Re: [Freesurfer] Matrix ill-conditioned error for dods thickness mri_glmfit analysis
posting the response to the list: The problem is that femalezero and femaletwo only have 4 and 3 subjects, respectively. As you are trying to estimate 4 slopes + an intercept, you will need at least 5 subjects per group. doug On 08/16/2017 03:27 PM, Sullivan, Danielle Renee wrote: > > Hi Doug, > > > I am just following up to this email. If you prefer I can repost on > the FreeSurfer mail archive. > > > Thanks, > > Danielle R. Sullivan (Miller), Ph.D. > > National Center for PTSD > > VA Boston Healthcare System Jamaica Plain > > Boston University School of Medicine > > OFFICE: (857) 364-4022 > > > > *From:* Sullivan, Danielle Renee > *Sent:* Friday, August 11, 2017 1:33:03 PM > *To:* Douglas N Greve > *Subject:* Re: [Freesurfer] Matrix ill-conditioned error for dods > thickness mri_glmfit analysis > > I am still getting the same error. My apologies for the back and forth. > > > Here are updated files (now there are 5 classes and contrast mtx has > been changed to reflect that). > > > Danielle R. Sullivan (Miller), Ph.D. > > National Center for PTSD > > VA Boston Healthcare System Jamaica Plain > > Boston University School of Medicine > > OFFICE: (857) 364-4022 > > > > *From:* Douglas N Greve <gr...@nmr.mgh.harvard.edu> > *Sent:* Friday, August 11, 2017 1:10:53 PM > *To:* Sullivan, Danielle Renee > *Subject:* Re: [Freesurfer] Matrix ill-conditioned error for dods > thickness mri_glmfit analysis > remove the definition too > > > On 08/11/2017 12:46 PM, Sullivan, Danielle Renee wrote: > > > > Hi, > > > > > > Thank you for the tips. I tried doing both and still run into the > > error (although now I get a warning that the class femaleone is > > defined but not used). Attached is my new FSGD (now all variables > > normalized via a Z-score), XG.mtx, Contrast, and Terminal output. > > > > > > Thanks again for the help! > > > > > > Danielle R. Sullivan (Miller), Ph.D. > > > > National Center for PTSD > > > > VA Boston Healthcare System Jamaica Plain > > > > Boston University School of Medicine > > > > OFFICE: (857) 364-4022 > > > > > > > > *From:* Douglas N Greve <gr...@nmr.mgh.harvard.edu> > > *Sent:* Friday, August 11, 2017 11:12:21 AM > > *To:* Sullivan, Danielle Renee > > *Subject:* Re: [Freesurfer] Matrix ill-conditioned error for dods > > thickness mri_glmfit analysis > > there appear to be two problems. One is that you only have one instance > > of femaleone. This might be ok if you had no covariates, but you can't > > estimate a covariate slope from one data point. Second, the scale of > > your covariates is much different, so you may need to normalize them. I > > would try removing the lone femaleone and rerun > > > > > > On 08/11/2017 10:35 AM, Sullivan, Danielle Renee wrote: > > > > > > Hi Doug, > > > > > > > > > Thanks for your response and help. > > > > > > > > > Attached is the matrix and the terminal output. Please let me know if > > > you need anything else. > > > > > > > > > > > > Danielle R. Sullivan (Miller), Ph.D. > > > > > > National Center for PTSD > > > > > > VA Boston Healthcare System Jamaica Plain > > > > > > Boston University School of Medicine > > > > > > OFFICE: (857) 364-4022 > > > > > > > > > can you send the design matrix Xg.mtx and the full terminal output? > > > On 08/08/2017 09:13 AM, Sullivan, Danielle Renee wrote: > > > > > > > > Hi FreeSurfer Experts, > > > > > > > > > > > > I am trying to run a dods cortical thickness analysis examining the > > > > interaction of group on a continuous variable while controlling for > > > > other variables. > > > > > > > > > > > > Specifically, I have 6 classes (gender by three groups [zero, one, > > > > two]). I am interested in the group interaction with a continuous > > > > variable while controlling for gender (among 4 other continuous > > > > variables). I set up my fsgd file (see attached) and contrast file > > (36 > > > > dimensions- F test): > > > > > >
Re: [Freesurfer] Matrix ill-conditioned error for dods thickness mri_glmfit analysis
can you send the design matrix Xg.mtx and the full terminal output? On 08/08/2017 09:13 AM, Sullivan, Danielle Renee wrote: > > Hi FreeSurfer Experts, > > > I am trying to run a dods cortical thickness analysis examining the > interaction of group on a continuous variable while controlling for > other variables. > > > Specifically, I have 6 classes (gender by three groups [zero, one, > two]). I am interested in the group interaction with a continuous > variable while controlling for gender (among 4 other continuous > variables). I set up my fsgd file (see attached) and contrast file (36 > dimensions- F test): > > *0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 [1 -1 0 1 > -1 0]/2* > > *0 0 0 0 0 0 **0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 [1 -0 > -1 1 0 -1]/2* > > > and ran this command: > > > *mri_glmfit --y > **lh.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forfsgd.thickness.10.mgh > > --fsgd > fsgd.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forFSGD.152.thickness.08082017.txt > > dods --C > /GLM_CONTRAST.MAT/Contrast.Male.Female.Age.PC1.PC2.PC3.GGT.RORAbySex.mtx > --surf fsaverage lh --cortex --glmdir > lh.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forfsgd.thickness.10.glmdir > > * > > > But keep getting this error: > > > *ERROR: matrix is ill-conditioned or badly scaled, condno = 1e+08* > > ** > > *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_glmfit --y > lh.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forfsgd.thickness.10.mgh > > --fsgd > fsgd.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forFSGD.152.thickness.08082017.txt > > dods --C > /Volumes/VA_Imaging/Projects/salat/2389_TRT/dmiller/GGT_CRP/GLM_CONTRAST.MAT/Contrast.Male.Female.Age.PC1.PC2.PC3.GGT.RORAbySex.mtx > > --surf fsaverage lh --cortex --glmdir > lh.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forfsgd.thickness.10.glmdir > > * > > * 2. The FSGD file (if using one)* > > * 3. And the design matrix above* > > > > Any idea on what I am doing wrong? > > > Danielle R. Miller, Ph.D. > > National Center for PTSD > > VA Boston Healthcare System Jamaica Plain > > Boston University School of Medicine > > OFFICE: (857) 364-4022 > > > > > ___ > 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 gr...@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] Matrix ill-conditioned error for dods thickness mri_glmfit analysis
Hi FreeSurfer Experts, I am trying to run a dods cortical thickness analysis examining the interaction of group on a continuous variable while controlling for other variables. Specifically, I have 6 classes (gender by three groups [zero, one, two]). I am interested in the group interaction with a continuous variable while controlling for gender (among 4 other continuous variables). I set up my fsgd file (see attached) and contrast file (36 dimensions- F test): 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 [1 -1 0 1 -1 0]/2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 [1 -0 -1 1 0 -1]/2 and ran this command: mri_glmfit --y lh.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forfsgd.thickness.10.mgh --fsgd fsgd.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forFSGD.152.thickness.08082017.txt dods --C /GLM_CONTRAST.MAT/Contrast.Male.Female.Age.PC1.PC2.PC3.GGT.RORAbySex.mtx --surf fsaverage lh --cortex --glmdir lh.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forfsgd.thickness.10.glmdir But keep getting this error: ERROR: matrix is ill-conditioned or badly scaled, condno = 1e+08 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_glmfit --y lh.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forfsgd.thickness.10.mgh --fsgd fsgd.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forFSGD.152.thickness.08082017.txt dods --C /Volumes/VA_Imaging/Projects/salat/2389_TRT/dmiller/GGT_CRP/GLM_CONTRAST.MAT/Contrast.Male.Female.Age.PC1.PC2.PC3.GGT.RORAbySex.mtx --surf fsaverage lh --cortex --glmdir lh.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forfsgd.thickness.10.glmdir 2. The FSGD file (if using one) 3. And the design matrix above Any idea on what I am doing wrong? Danielle R. Miller, Ph.D. National Center for PTSD VA Boston Healthcare System Jamaica Plain Boston University School of Medicine OFFICE: (857) 364-4022 GroupDescriptorFile 1 Title fsgd.age.PC1_forFSGD.PC2_forFSGD.PC3_forFSGD.GGT_forFSGD.rs8042149_A_bysex_forFSGD.152.thickness.08082017.txt MeasurementName thickness Class femalezero Class femaleone Class femaletwo Class malezero Class maleone Class maletwo Variables age PC1_forFSGD PC2_forFSGD PC3_forFSGD GGT_forFSGD #Input TRT_0007 = 43 = = = = #Input TRT_0015 = 38 = = = = #Input TRT_0016 femalezero 47 0.0295 -0.1271 0.0693 = #Input TRT_0017 maleone 29 0.0583 0.1283 -0.0139 = #Input TRT_0019 = 43 = = = = #Input TRT_0020 = 28 = = = = #Input TRT_0022 malezero 23 0.0324 0.0381 0.0355 = #Input TRT_0025 maletwo 24 -0.0619 -0.0585 -0.0010 = #Input TRT_0026 maleone 26 0.0344 0.1225 0.0991 = #Input TRT_0027 femaletwo 43 -0.0417 0.0989 0.0776 = #Input TRT_0028 = 37 = = = = #Input TRT_0030 maleone 23 0.0230 0.0637 0.0496 = #Input TRT_0031 maleone 19 -0.0498 0.0925 -0.0851 = #Input TRT_0032 femalezero 25 -0.0151 -0.0589 -0.0538 = #Input TRT_0033 malezero 24 0.0675 -0.1426 0.0793 = #Input TRT_0034 maletwo 39 -0.0826 0.0282 0.0109 = #Input TRT_0035 maletwo 32 0.1583 -0.0853 -0.0525 = #Input TRT_0036 maleone 49 0.1116 -0.0184 -0.0206 = #Input TRT_0037 malezero 27 0.0836 0.0867 0.0418 = #Input TRT_0038 = 44 = = = = Input TRT_0042 malezero 20 0.0775 0.0252 0.0351 15 #Input TRT_0043 maleone 49 -0.0615 -0.0789 0.0855 = #Input TRT_0045 malezero 38 -0.0078 -0.0248 -0.0194 = #Input TRT_0047 maletwo 26 -0.0431 -0.0585 -0.0031 = #Input TRT_0048 = 31 = = = 28 #Input TRT_0049 femaletwo 23 -0.0237 0.0189 -0.0160 = #Input TRT_0050 = 32 = = = = #Input TRT_0051 = 43 = = = = #Input TRT_0052 malezero 29 -0.0980 0.0265 0.1271 = #Input TRT_0055 malezero 33 0.1757 0.0140 -0.0469 = Input TRT_0057 maletwo 26 0.1976 0.0419 -0.0359 32 #Input TRT_0058 = 58 = = = 10 Input TRT_0059 malezero 42 -0.0061 0.1113 -0.0729 29 #Input TRT_0060 = 37 = = = 33 Input TRT_0061 maleone 34 0.0508 -0.0194 0.0931 22 Input TRT_0062 maletwo 42 -0.1008 -0.1552 0.0298 46 #Input TRT_0063 maleone 46 -0.0077 -0.0068 0.0540 = Input TRT_0064 maleone 47 0.0792 -0.1128 0.1746 13 Input TRT_0065 maletwo 55 -0.0489 -0.1078 0.0532 65 #Input TRT_0066 = 35 = = = 26 Input TRT_0067 maleone 40 -0.0759 0.0753 -0.0375 24 Input TRT_0068 maleone 24 -0.1071 0.0256 -0.0220 48 #Input TRT_0069 = 44 = = = 26 Input TRT_0070 maleone 47 -0.0551 0.0043 0.0556 29 Input TRT_0071 malezero 28 0.0037 -0.0156 0.0113 19 Input TRT_0072 malezero 31 0.0855 -0.0362 -0.0036 20 Input TRT_0073 malezero 21 -0.0363 0.0048 0.0014 56 Input TRT_0075 maletwo 42 -0.0713 0.0286 -0.1274 30 Input TRT_0076 malezero 33 0.0995 -0.0306 0.0634 23 Input TRT_0078 maleone 39 0.1166 -0.1302 -0.0220 19 Input TRT_0079 maletwo 33 -0.0696 0.0533 -0.0207 24 Input TRT_0082 malezero 58 0.0669 -0.0484 -0.1014 54 Input
