Hi again, let me reiterate how helpful everyone has been. I really appreciate it and I have been able to create a 360x360 correlation matrix for test sample of 10 HCP subjects. I was looking at the wb_command -cifti-correlation and noticed you can make a covariance matrix, and I was wondering what method you recommend using to get the inverse covariance matrix? I read Smith et al 2011, where they recommend http://www.cs.ubc.ca/~schmidtm/Software/L1precision.html.<http://www.cs.ubc.ca/~schmidtm/Software/L1precision.html> I have also seen fslNETs used (which I believe uses the same L1 precision). I read that Dr. Harms suggested and provided input on how to use fslNETs http://www.mail-archive.com/[email protected]/msg01341.html (as well as suggested on this thread to compute a separate matrix for each subject). Is there a way to input a correlation or covariance matrix into fslNETs? (this might be a question of the fsl list serv)
Thank you again, Grace Grace Shearrer, PhD Post Doctoral Researcher Neuropsychology of Ingestive Behavior Lab University of North Carolina Chapel Hill ________________________________ From: Glasser, Matthew <[email protected]> Sent: Sunday, January 8, 2017 1:09 PM To: Shearrer, Grace Cc: Jennifer Elam; NEUROSCIENCE tim Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI Hi Grace, Have a look at wb_command -cifti-help Matt. From: "Shearrer, Grace" <[email protected]<mailto:[email protected]>> Date: Sunday, January 8, 2017 at 11:06 AM To: Matt Glasser <[email protected]<mailto:[email protected]>>, Timothy Coalson <[email protected]<mailto:[email protected]>> Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI Hi TIm and Matt, thank you so much for your help again. This might be a silly question, but is there any documentation on the different file types (eg .dseries.nii vs .dscalar.nii vs .ptseries.nii)? As I am doing my analysis I am trying to understand each step in a thoughtful manner so hopefully I can teach grad students in my lab in the future. I searched the HCP documentation but couldn't find anything about the different file types, I may have missed it though. Thank you, Grace Grace Shearrer, PhD Post Doctoral Researcher Neuropsychology of Ingestive Behavior Lab University of North Carolina Chapel Hill ________________________________ From: Glasser, Matthew <[email protected]<mailto:[email protected]>> Sent: Friday, January 6, 2017 9:35 PM To: NEUROSCIENCE tim Cc: Shearrer, Grace Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI Thanks Tim. Matt. From: Timothy Coalson <[email protected]<mailto:[email protected]>> Date: Friday, January 6, 2017 at 7:33 PM To: Matt Glasser <[email protected]<mailto:[email protected]>> Cc: "Shearrer, Grace" <[email protected]<mailto:[email protected]>> Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI Well, I wrote it as an option to the -*-concatenate functions, and doing so has warned me not to do it that way again in the future - the script file has almost doubled in size, as each of the one-line loops in those commands is now 50 lines (error checking, avoiding reading a data file into shell memory if the user mistypes things...). I also didn't add it to -cifti-demean (that could be a bit messier). You can try it by getting the latest version here: https://github.com/Washington-University/wb_shortcuts There is equivalent shell magic to do it (with whitespace safety) without using this option, but it will choke on empty lines in the text file: readarray -t filelist < filelist.txt wb_shortcuts -cifti-concatenate output.dtseries.nii "${filelist[@]}" If someone asks how to do it for -cifti-demean, I think this shell incantation may be the better answer. Tim On Fri, Jan 6, 2017 at 4:30 PM, Glasser, Matthew <[email protected]<mailto:[email protected]>> wrote: Whatever is most convenient. Matt. From: Timothy Coalson <[email protected]<mailto:[email protected]>> Date: Friday, January 6, 2017 at 5:29 PM To: Matt Glasser <[email protected]<mailto:[email protected]>> Cc: "Shearrer, Grace" <[email protected]<mailto:[email protected]>> Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI It might be able to fit into the -*-concatenate parsing (people shouldn't name files starting with '-'). It could also be a separate shortcut. Tim On Fri, Jan 6, 2017 at 4:10 PM, Glasser, Matthew <[email protected]<mailto:[email protected]>> wrote: That functionality might be a good candidate for wb_shortcuts. If you are merging that many subjects, you might consider how large your file will be. You might want to make use of an algorithm like MIGP to do the combination. Matt. From: "Shearrer, Grace" <[email protected]<mailto:[email protected]>> Date: Friday, January 6, 2017 at 11:57 AM To: Timothy Coalson <[email protected]<mailto:[email protected]>> Cc: Matt Glasser <[email protected]<mailto:[email protected]>>, "Elam, Jennifer" <[email protected]<mailto:[email protected]>>, "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>>, "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI Hi Tim, I have a follow up question about the cifti-merge command. Given I will be merging upwards of 500 subjects does the cifti-merge command have the ability to take a .txt listing the paths to all the parcellated files? Similar to the fsl's group ica function which allows: melodic -i inputlist_clean.txt -o groupmelodic_fix.ica -v --nobet --bgthreshold=10 --tr=2.000 --report -d 25 --mmthresh=0.5 --Ostats -a concat Thanks, Grace ________________________________ From: Timothy Coalson <[email protected]<mailto:[email protected]>> Sent: Wednesday, January 4, 2017 8:43 PM To: Shearrer, Grace Cc: Glasser, Matthew; Elam, Jennifer; [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]> Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI Not quite, you still need to do -cifti-merge to concatenate across subjects to get one matrix that reflects data from all subjects. Per subject, with a little bit of pattern matching magic: 1) wb_shortcuts -cifti-demean -normalize <subject-output-dir>/all_normalized.dtseries.nii rfMRI_REST?*/rfMRI_REST*Atlas.dtseries.nii 2) wb_command -cifti-parcellate Once, across all subjects: 3) wb_command -cifti-merge (or wb_shortcuts -cifti-concatenate, which is a friendlier front-end to the same command - due to a bug, you'd need the latest version, here: https://github.com/Washington-University/wb_shortcuts/raw/master/wb_shortcuts) 4) wb_command -cifti-correlation Tim On Wed, Jan 4, 2017 at 4:09 PM, Shearrer, Grace <[email protected]<mailto:[email protected]>> wrote: Hi Tim, yes I am planning on concatenating all subjects. Okay so new work flow to make sure I understand: 1) wb_shortcuts -cifti-demean -normalize output.dtseries.nii input.dtseries.nii #do this on each subject where the input would be the 4 resting states runs 2) wb_command -cifti-parcellate #on each subject 3) wb_command -cifit-correlation #this will create a 360x360 matrix for further analysis Thanks so much for tips. I am brand new to workbench so I need all the help I can get. Grace ________________________________ From: Timothy Coalson <[email protected]<mailto:[email protected]>> Sent: Wednesday, January 4, 2017 7:00 PM To: Shearrer, Grace Cc: Glasser, Matthew; Elam, Jennifer; [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]> Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI Are you planning to concatenate the timeseries across all subjects? If so, I would suggest parcellating in each subject before concatenating (so, swap steps 4 and 5), so that the merged file isn't enormous, and so that the parcellation can be done in parallel on a cluster. The result of concatenating before vs after parcellation should be the same if you use the same arguments to -cifti-parcellate, but on the other hand, parcellating in each subject would make it possible to use individual subject parcellations instead of a group parcellation. Side note, wb_shortcuts has a -cifti-demean function that can do steps 1 through 3 for you (assuming you don't want to keep the mean and stdev dense maps around - also note that the mean and stdev files should actually be dscalar.nii files, not dtseries.nii): wb_shortcuts -cifti-demean -normalize output.dtseries.nii input.dtseries.nii This command can also merge files, so you could for instance merge the 4 runs of each subject after normalizing them by specifying all of them as inputs to that one command, and then only need to do -cifti-parcellate on one file per subject. There is also some explanatory text for the cifti format in the information options of wb_command: http://www.humanconnectome.org/software/workbench-command.php?function=-cifti-help Tim On Wed, Jan 4, 2017 at 3:28 PM, Shearrer, Grace <[email protected]<mailto:[email protected]>> wrote: Ah! Most excellent. Thank you again for your prompt reply. 1) wb_command -cifti-reduce <input> MEAN mean.dtseries.nii 2) wb_command -cifti-reduce <input> STDEV stdev.dtseries.nii 3) wb_command -cifti-math '(x - mean) / stdev' <output> -fixnan 0 -var x <input> -var mean mean.dtseries.nii -select 1 1 -repeat -var stdev stdev.dtseries.nii -select 1 1 -repeat 4) wb_command -cifit-merge 5) wb_command -cifti-parcellate 6) wb_command -cifit-correlation Grace ________________________________ From: Glasser, Matthew <[email protected]<mailto:[email protected]>> Sent: Wednesday, January 4, 2017 6:22 PM To: Shearrer, Grace; Elam, Jennifer; [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]> Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI I would insert the parcellation between steps 5 and 6. The result if you use the file I suggested below will be a 360 X 360 cortical area correlation matrix. Peace, Matt. From: "Shearrer, Grace" <[email protected]<mailto:[email protected]>> Date: Wednesday, January 4, 2017 at 4:16 PM To: Matt Glasser <[email protected]<mailto:[email protected]>>, "Elam, Jennifer" <[email protected]<mailto:[email protected]>>, "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>>, "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI Hi Matt, thanks so much for all your work on HCP and for replying to my email, I really appreciate it (thank you to Jenn too!). I want to make sure I understand the sequence quickly: 1) wb_command -cifti-parcellate 2) wb_command -cifti-reduce <input> MEAN mean.dtseries.nii 3) wb_command -cifti-reduce <input> STDEV stdev.dtseries.nii 4) wb_command -cifti-math '(x - mean) / stdev' <output> -fixnan 0 -var x <input> -var mean mean.dtseries.nii -select 1 1 -repeat -var stdev stdev.dtseries.nii -select 1 1 -repeat 5) wb_command -cifit-merge 6) wb_command -cifit-correlation And then I would take this (giant) matrix and run my analyses on it with the brain connectivity toolbox (or the like). Thank you for your patience, Grace ________________________________ From: Glasser, Matthew <[email protected]<mailto:[email protected]>> Sent: Wednesday, January 4, 2017 5:54 PM To: Elam, Jennifer; Shearrer, Grace; [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]> Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI You will need to parcellate the data first using wb_command -cifti-parcellate. A parcellation is available here: https://balsa.wustl.edu/file/show/3VLx Peace, Matt. From: <[email protected]<mailto:[email protected]>> on behalf of "Elam, Jennifer" <[email protected]<mailto:[email protected]>> Date: Wednesday, January 4, 2017 at 3:47 PM To: "Shearrer, Grace" <[email protected]<mailto:[email protected]>>, "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>>, "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI Hi Grace, This is a grayordinate x grayordinate "dense" connectome and it will be a large (30GB) resulting file. Grayordinates are surface vertices for both hemispheres + subcortical voxels (total of 91,282 grayordinates in our case). To learn more about this and CIFTI, I would look at Glasser et al. 2013.<http://www.sciencedirect.com/science/article/pii/S1053811913005053> <http://www.sciencedirect.com/science/article/pii/S1053811913005053> Best, Jenn Jennifer Elam, Ph.D. Scientific Outreach, Human Connectome Project Washington University School of Medicine Department of Neuroscience, Box 8108 660 South Euclid Avenue St. Louis, MO 63110 314-362-9387<tel:314-362-9387> [email protected]<mailto:[email protected]> www.humanconnectome.org<http://www.humanconnectome.org/> ________________________________ From: Shearrer, Grace <[email protected]<mailto:[email protected]>> Sent: Wednesday, January 4, 2017 2:16:48 PM To: Elam, Jennifer; [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]> Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI Hi Jenn, oops the User list verification was going to my junk email. I just signed up though. I did see the wiki and it was helpful, I like this script though since it appears to loop through all the directories. I just wanted to make sure I was getting the final connectivity matrix. One other question, is the connectivity matrix voxelwise? Or is it connectivity based on a parcelation scheme? Thank you, Grace Grace Shearrer, PhD Post Doctoral Researcher Neuropsychology of Ingestive Behavior Lab University of North Carolina Chapel Hill ________________________________ From: Elam, Jennifer <[email protected]<mailto:[email protected]>> Sent: Wednesday, January 4, 2017 4:57 PM To: Shearrer, Grace; [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]> Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI Hi Grace, You might also want to check out the instructions for creating individual and group average dense rfMRI connectomes in item #3 in the HCP Users FAQ on our wiki: https://wiki.humanconnectome.org/display/PublicData/HCP+Users+FAQ#HCPUsersFAQ-3.IseethatHCPdistributesgroupaveragedenseconnectomefiles.Doyoualsoprovideconnectivitymatricesforindividualsubjects? <https://wiki.humanconnectome.org/display/PublicData/HCP+Users+FAQ#HCPUsersFAQ-3.IseethatHCPdistributesgroupaveragedenseconnectomefiles.Doyoualsoprovideconnectivitymatricesforindividualsubjects?> Also, sign up for the HCP-Users list<https://www.humanconnectome.org/contact/#subscribe> to learn from other users and stay in the discussion. Best, Jenn Jennifer Elam, Ph.D. Scientific Outreach, Human Connectome Project Washington University School of Medicine Department of Neuroscience, Box 8108 660 South Euclid Avenue St. Louis, MO 63110 314-362-9387<tel:314-362-9387> [email protected]<mailto:[email protected]> www.humanconnectome.org<http://www.humanconnectome.org/> ________________________________ From:[email protected]<mailto:[email protected]> <[email protected]<mailto:[email protected]>> on behalf of Shearrer, Grace <[email protected]<mailto:[email protected]>> Sent: Wednesday, January 4, 2017 1:24:49 PM To: [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]> Subject: Re: [HCP-Users] Creating Group Average Resting State fMRI Hi Dr. Bari, thank you so much for script this is very helpful. I just want to check that I understand. I would use your script to get the mean time series, standard deviation, and then demean the data. Then I would use a command like: wb_command -cifit-merge to then merge all my demeaned data into a a single file, then I would use: wb_command -cifit-correlation to get a correlation matrix from which I could begin to use a graph theory toolkit (like brain connectivity tool box). Does this make sense? This is my first time using CIFTI files and I am still trying to wrap my head around it. Thanks, Grace This question is based on the following thread: http://www.mail-archive.com/[email protected]/msg00448.html Grace Shearrer, PhD Post Doctoral Researcher Neuropsychology of Ingestive Behavior Lab University of North Carolina Chapel Hill _______________________________________________ HCP-Users mailing list [email protected]<mailto:[email protected]> http://lists.humanconnectome.org/mailman/listinfo/hcp-users _______________________________________________ HCP-Users mailing list [email protected]<mailto:[email protected]> http://lists.humanconnectome.org/mailman/listinfo/hcp-users ________________________________ The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. ________________________________ The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. _______________________________________________ HCP-Users mailing list [email protected]<mailto:[email protected]> http://lists.humanconnectome.org/mailman/listinfo/hcp-users ________________________________ The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. ________________________________ The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. ________________________________ The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. ________________________________ The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. _______________________________________________ HCP-Users mailing list [email protected] http://lists.humanconnectome.org/mailman/listinfo/hcp-users
