External Email - Use Caution
Hi, Doug. I manually checked for the sig values, and all I got were positive. Is this unexpected?
Should I write someone to get those binaries? I tried to run it again, and was not successful.
Many many thanks!
-- Pedro
On May 29, 2018, at 12:29, Douglas Greve dgreve@mgh.harvard.edu wrote:
On 5/27/18 3:33 PM, Pedro Rosa wrote:
Hi, Doug. For this specific 'problematic' cluster, the maximum statistical value from the cluster summary is negative, that indicates a direction of result (considering the contrast matrix I used) opposite to the one I get when I compare means. I inspected p-values from sig.mgh (and also p-values masked for surviving clusters) and all seem positive. I could not actually get how inspection sig.mgh may help to solve this issue. Can you clarify this point, please?
I mean do you find both positive and negative values in the sig.mgh in the vertices of the problematic cluster? I suppose it is possible, I've just never seen it before.
Following your comment on the 1.3 threshold, I read the paper and the permutation page from FreeSurfer wiki, and tried to run mri_glmfit-sim with perm function (after mri_glmfit with -eres function) using FS6 twice, and it seems to take many hours (~ 26 hours) to run (I am using a regular Mac computer, not a cluster), while the wiki mentions it should take 20 minutes. I got this message while running it: 'INFO: design matrix is not orthogonal, but perm forced'. Also, it ended with this error: ERROR: Option --vwsigmax unknown., and it dit not creat the perm* files. I attached the log, in case you want to inspect it.
If you are on a mac, we will have to get new binaries for you. Rob, can you get a mac version of mri_surfcluster to Pedro?
Also, don't run mri_glmfit-sim with --bg 1 as it causes really long run times for reasons I don't understand. If you are running with multiple processors, then use --bg N where N is the number, but if N=1, just leave off --bg entirely.
Also, I changed the mri_glmfit-sim file regarding the .csh file path ($FREESURFER_HOME/sources.csh to /SetUpFreeSurfer.csh) before running it (it could not find sources.csh). Many thanks, Pedro.
On Wed, May 23, 2018 at 6:45 PM, Douglas N. Greve dgreve@mgh.harvard.edu wrote: So the maximum is positive, but the mean is negative? When you look at the uncorrected p-values, can you actually find this? My guess is that the cluster must be largely negative, so I would report that
On 05/23/2018 03:52 PM, Pedro Rosa wrote:
External Email - Use CautionDear Doug, Many thanks. I ran the analysis with 'abs', 'pos' and 'neg' testing. All clusters (including the 'problematic' cluster on the medialorbitofrontal region) are present in 'abs' and 'neg' testing summaries, and not present in 'pos' testing summary. This medialorbitofrontal cluster shows a direction of effect when I consider the contrast (and the maximum statistical value signal) and the opposite direction when I analyze groups means. How should I proceed to report such result? Best, Pedro. PS: I apologize if I sent a number of empty emails to the list in the last hours.
On May 23, 2018, at 13:30, pedrogomesrosa@gmail.com wrote:
Dear Doug
-- Pedro
On May 15, 2018, at 13:36, Douglas N. Greve dgreve@mgh.harvard.edu wrote:
depends on what you are getting the direction of the result for and how you want to interpret it. If the cluster has both pos and neg, then the interpretation becomes more difficult. Is this really happening? I don't think I've ever seen it. The abs() just takes the absolute value, there is no summing across the cluster, just counting voxel above threshold. If you really want to evaluate the sign, you can do a signed test (corrections across signed and unsigned tests then become problematic). Finally, the 1.3 threshold is way to low when using the MC-Z simulation. If you want to use a threshold that low, then you'll need to use a permutation instead. See https://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/MultipleComparisonsV6.0...
> On 05/14/2018 06:33 PM, Pedro Rosa wrote: > > > Many thanks, Doug! > Should I then rely on groups' means to determine the 'direction of the > result', and not on the 'max statistical significance' signal? > Does 'abs' simulation sum up the results I could get by running 'pos' > and 'neg' simulations separately? > Best > > On Mon, May 14, 2018 at 7:13 PM, Douglas N. Greve > <dgreve@mgh.harvard.edu mailto:dgreve@mgh.harvard.edu> wrote: > > You have specified "abs" meaning an unsigned "absolute" test, so > there > is no guarantee that the mean contrast will be pos or neg. > > >> On 05/14/2018 06:11 PM, Pedro Rosa wrote: >> >> >> Here it is: mri_glmfit-sim --glmdir >> PEP-HC_Gender_Age.txt.lh.area.mgh.5.mgh.cortex.glmdir --sim-sign > abs >> --cache 1.3 abs >> >> On Mon, May 14, 2018 at 6:59 PM, Douglas N. Greve >> <dgreve@mgh.harvard.edu mailto:dgreve@mgh.harvard.edu > <mailto:dgreve@mgh.harvard.edu mailto:dgreve@mgh.harvard.edu>> > wrote: >> sorry, one more thing. Can you send your mri_glmfit-sim > command line? >> >> On 05/14/2018 05:46 PM, Pedro Rosa wrote: >> > >> > >> > Hi, Doug. Many thanks. >> > Find the fsgd and the contrast attached, as well as the > cluster >> > summary and the individual-based cluster data (which I used to >> > calculate the groups' means). The cluster I referred to is the >> second >> > one (medialorbitofrontal). >> > I investigated differences in cortical area from groups 1 > and 2 in >> > contrast to groups 3 and 4 (contrast 3 as in >> > https://surfer.nmr.mgh.harvard.edu/fswiki/Fsgdf4G1V > https://surfer.nmr.mgh.harvard.edu/fswiki/Fsgdf4G1V >> https://surfer.nmr.mgh.harvard.edu/fswiki/Fsgdf4G1V https://surfer.nmr.mgh.harvard.edu/fswiki/Fsgdf4G1V>). >> > Best, >> > Pedro Rosa. >> > >> > On Mon, May 14, 2018 at 5:42 PM, Douglas N. Greve >> > <dgreve@mgh.harvard.edu mailto:dgreve@mgh.harvard.edu > <mailto:dgreve@mgh.harvard.edu mailto:dgreve@mgh.harvard.edu> >> mailto:dgreve@mgh.harvard.edu mailto:dgreve@mgh.harvard.edu mailto:dgreve@mgh.harvard.edu mailto:dgreve@mgh.harvard.edu>>> >> wrote: >> > >> > Can you send your fsgd file and contrasts? >> > >> > >> > On 05/14/2018 01:25 PM, Pedro Rosa wrote: >> > > >> > > >> > > Dear FreeSurfers, >> > > I got results from a cortical area analysis that > intrigued >> me: Upon >> > > the inspection of clusters that survived Monte-Carlo >> simulation, I >> > > found a cluster with larger area among a first group in >> contrast >> > to a >> > > second group. >> > > However, when I calculated the cortical areas means > from that >> > cluster >> > > (as available in the glmdir), the second group had > values >> slightly >> > > larger than the first group (i.e., showing a > difference in the >> > > opposite direction). >> > > How would you explain such findings? Must I have > committed a >> > mistake >> > > in the analysis or its interpretation? >> > > Best, >> > > Pedro Rosa - University of São Paulo - Brazil. >> > > >> > > >> > > _______________________________________________ >> > > Freesurfer mailing list >> > > Freesurfer@nmr.mgh.harvard.edu > mailto:Freesurfer@nmr.mgh.harvard.edu >> mailto:Freesurfer@nmr.mgh.harvard.edu mailto:Freesurfer@nmr.mgh.harvard.edu> >> > mailto:Freesurfer@nmr.mgh.harvard.edu mailto:Freesurfer@nmr.mgh.harvard.edu >> mailto:Freesurfer@nmr.mgh.harvard.edu mailto:Freesurfer@nmr.mgh.harvard.edu>> >> > > >> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer> >> > > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer>> >> > >> > _______________________________________________ >> > Freesurfer mailing list >> > Freesurfer@nmr.mgh.harvard.edu > mailto:Freesurfer@nmr.mgh.harvard.edu >> mailto:Freesurfer@nmr.mgh.harvard.edu mailto:Freesurfer@nmr.mgh.harvard.edu> >> mailto:Freesurfer@nmr.mgh.harvard.edu mailto:Freesurfer@nmr.mgh.harvard.edu >> mailto:Freesurfer@nmr.mgh.harvard.edu mailto:Freesurfer@nmr.mgh.harvard.edu>> >> > > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer> >> > >> > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer 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 > http://www.partners.org/complianceline >> http://www.partners.org/complianceline http://www.partners.org/complianceline> >> > http://www.partners.org/complianceline http://www.partners.org/complianceline >> http://www.partners.org/complianceline 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 > mailto:Freesurfer@nmr.mgh.harvard.edu >> mailto:Freesurfer@nmr.mgh.harvard.edu mailto:Freesurfer@nmr.mgh.harvard.edu> >> > > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer> >> _______________________________________________ >> Freesurfer mailing list >> Freesurfer@nmr.mgh.harvard.edu > mailto:Freesurfer@nmr.mgh.harvard.edu > mailto:Freesurfer@nmr.mgh.harvard.edu mailto:Freesurfer@nmr.mgh.harvard.edu> >> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer> >> >> >> >> _______________________________________________ >> Freesurfer mailing list >> Freesurfer@nmr.mgh.harvard.edu > mailto:Freesurfer@nmr.mgh.harvard.edu >> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > > _______________________________________________ > Freesurfer mailing list > Freesurfer@nmr.mgh.harvard.edu mailto:Freesurfer@nmr.mgh.harvard.edu > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > 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 _______________________________________________ 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
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
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.