| T O P I C R E V I E W |
| Dichter |
Posted - Jun 20 2007 : 10:22:21 AM
I am trying to use bxh_eventstats_combine to analyze data from subjects in two diagnostic groups (patient control), and both groups are scanned twice (time1, time2). To accomplish this, I am assigning every scan to one of four conditions: control1, control2, patient1, patient2
So my useprefixfile contains iterations of the following:
#prefix group
/home/dichter/net/goldman/data/BIAC/DepTher.02/Analysis/eventstat/46180/stats control1
etc....
Two questions:
(1) My output contains a "supergroup" that is the average of all subjects. However, in addition I would like to look at a "supergroup" that is averaged over time1 and time 2, but not over group.
(2) My output folder structure is a little odd -- I want to make sure I have things set up correctly: within the top level output folder, there is a single folder called 'group' but no other data files; within that folder are the four folder of interest: depressed1
depressed2, control1, control2. IS that right?
Thank you,
-Gabriel
|
| 13 L A T E S T R E P L I E S (Newest First) |
| Dichter |
Posted - Jul 12 2007 : 12:51:37 PM
bxh_eventstats_combine is completing without errors, but the output only includes "_n" and "_avg" files (no _tmaps or _avg_percent files are created). Thanks for any help.
-Gabriel
|
| syam.gadde |
Posted - Jul 03 2007 : 2:17:05 PM
The /CIFS/Goldman directory structure is specific to golgi, and how it was set up. Your particular machine might have a different set up and the path names will have to be adjusted accordingly. |
| Dichter |
Posted - Jul 03 2007 : 2:05:11 PM
It works of Golgi, but not our Linux box (Meehl). I guess /CIFS/Goldman/Data won't work in other OS environments? Thanks. |
| syam.gadde |
Posted - Jul 03 2007 : 09:56:43 AM
Did you misspell the directory name eventstat (instead of eventstats)? |
| Dichter |
Posted - Jul 02 2007 : 10:38:58 PM
For a seperate study, the /CIFS/Goldman/Data prefix appears to no longer be working, e.g., using the following path
/CIFS/Goldman/Data/NDRC/Cpram.02/Analysis/eventstat/33216/stats control time1 controltime1
gives the error:
Use of uninitialized value in concatenation (.) or string at /usr/local/bin/bxh_eventstats_combine line 261.
Error opening '/CIFS/Goldman/Data/NDRC/Cpram.02/Analysis/eventstat/33216/stats_LOG.txt': No such file or directory |
| syam.gadde |
Posted - Jun 21 2007 : 12:11:28 PM
There is no --overwrite option at the moment. I generally don't like to provide those because they tend to become default, though I make exceptions if it is something that gets done a lot, or it is cumbersome to delete the output. In this case it is easy enough since there is only one output directory, and you can use "rm -r" (be careful with that). |
| Dichter |
Posted - Jun 21 2007 : 11:55:11 AM
Thank you, that worked.
Is there a way to allow overwriting of output? |
| syam.gadde |
Posted - Jun 21 2007 : 11:34:04 AM
That doesn't work, unfortunately. The ~ character is expanded by the shell but not by most programs. You could try /CIFS/Goldman/Data which is user independent... |
| Dichter |
Posted - Jun 21 2007 : 11:29:44 AM
I am trying to modify my useprefixfile file so that it is user-independent. It works correctly by pointing to scans like this:
/home/dichter/net/goldman/data/BIAC/DepTher.02/Analysis/eventstat/46180/stats patient time1 patienttime1
However, not like this:
~/net/goldman/data/BIAC/DepTher.02/Analysis/eventstat/46180/stats depressed1
Thanks for any help. |
| Dichter |
Posted - Jun 20 2007 : 1:25:04 PM
That's perfect. Thank you. |
| syam.gadde |
Posted - Jun 20 2007 : 12:56:04 PM
You should be able to do this to get all combinations:
# prefix group visit groupvisit
prefix1 control visit1 controlvisit1
prefix2 control visit2 controlvisit2
prefix3 patient visit1 patientvisit1
prefix4 patient visit2 patientvisit2
prefix5 patient visit1 patientvisit1
prefix6 patient visit2 patientvisit2
prefix7 control visit1 controlvisit1
prefix8 control visit2 controlvisit2
The group directory would have statistics separated by control and patient (i.e. 2 directories), the visit directory should have them separated by visit number (i.e. 2 directories), and groupvisit would have 4 directories. Each of the three "axes" covers all the input data, just divided into groups differently. I assume that will do what you need?
|
| Dichter |
Posted - Jun 20 2007 : 12:00:37 PM
Thanks. When I use a second column called "visit", however, I have the inverse problem, if you will. In addition to the supergroup, there are folders for:
patients collapsed across time
controls collapsed across time
time 1 collapsed across groups
time 2 collapsed across groups
However, I never get seperate analyses of control time 1, control time 2, etc.
In other words, the strategy I mentioned in my initial post allows for examinations of individual levels of groupXtime, but not groups collapsed across time. The second strategy allows for collapsing across levels of each factor, but not specifc cells within the groupXtime design. Is there a way to do all of the above in a single sweep? Thanks. |
| syam.gadde |
Posted - Jun 20 2007 : 10:32:09 AM
Within the top-level output directory, the images represent the supergroup that contains all inputs. There is also a PREFIX_groups directory, within which there should be a directory for each of the columns you have in your prefix file (you would have a "group" directory), within which you will see the various divisions that you specified for that column. So it looks correct.
If you want another set of groups that separates the inputs by the visit number, you can create another column called "visit" and label each input with "visit1" or "visit2" (or whatever labels you want to use).
Then you should find a "group" and "visit" directory in the PREFIX_groups output directory. |
