Hi Tracula team,
I have just successfully processed my first scan through all the Tracula steps! Though, I have to say it could have gone much smoother / quicker if the documentation on the wiki were a little more up-to-date. I realize it's an ongoing struggle to keep documentation current, so I thought I'd send you my observations as I worked through all the bumps in the road. You are probably aware of most of these issues but here they are in one big email for you to consider:
1. I'm unable to find anything among the Tracula setup instructions where it says which version of FSL that Tracula requires, and this version issue was a crucial piece of the puzzle. I had been failing at step 2 with an older version of FSL (v4.1.4), and per your suggestion I updated to v4.1.9. However, when called by Trac-all -bedp, bedpostx ALWAYS failed to created the "merged*.nii.gz" files for me, even after modifying it and updating FSL versions. I'm not sure why that is but it may be helpful to users to know they can simply bypass that step by launching bedpostx separately in the latest FSL, and pointing it at the "dmri" folder created during the Trac-all -prep stage. Also, we had to modify "bedpostx", as well as "fsl_sub", to get it to launch to our cluster without issues. Once all the fixes were in place, this step took only an hour or two to run. I then copied bedpostx's outputs into the requisite "dmri.bedpostX" directory to launch the -path step, and it completed in about an hour. Launching "bedpostx_datacheck <dmri folder>" beforehand was very helpful for troubleshooting the prep stage.
2. There is a typo in the description of the configuration file setup: "set dorotvecs" is listed in the detailed description under Step9, but this is incorrect and will cause errors. It's a bit confusing because it is correctly listed as "set dorotbvecs" in the complete config file example you provide above that section. It's unclear which one is correct and forces users to debug something which they probably have never run before.
3. Also in the config file is the variable $FSLDIR, which I discovered we had to set after receiving errors during mni registration. I suppose the setting of that variable should be obvious, but it wasn't clear based on the example which folder, exactly, should be associated with that variable name. We set our $FSLDIR to the /bin folder and all was well. Entirely possible this is just a lab-specific issue on our end. Since this variable needs to be set properly, just like the $SUBJECTS_DIR, perhaps this information could be included towards the top of the page where the rest of the required environmental variables are mentioned?
4. Finally, in the "Tractography Outputs" section, the command line example given needs to be updated because it shows "lh.ilf_AS_avg32_mni_flt" as example. The "32" should be "33". The same goes for "merged_avg32_mni_flt.mgz", obviously.
Until very recently, I was experiencing all the same issues Scott Hayes has been reporting lately. (The singular "Data_slice_0002" folder, with missing files, and the over-and-over messages of "1 slices processed" in a terminal that never throws an error message despite its failure. During the subsequent -path step, without those "merged*.nii.gz" files, I would receive the dreaded & perplexing "Word Too Long" error. A more specific error message regarding the missing files might be more useful for troubleshooting purposes.)
Many thanks for all your advice and help throughout the process! We can now move forward and do some exploration... Very exciting!
Shannon T. Buckley SRA III - Supervisor CIND, UCSF 4150 Clement St. (114M) San Francisco, CA 94121 Phone: (415) 221-4810 x4251 Fax: (415) 386-3954