Running on concatenated runs

Is there an option to run on concatenated (and pre-processed) runs and define the start points of each run, similar to the -concat option for 3dDeconvolve?

I want to run a new model on already-processed data, and there seems to be no need to re-run all the pre-processing blocks from the raw data, as I have the all_runs data saved. I know I could just run 3dDeconvolve and then run the REML output script, but the processing pipeline would be more straight forward if I could run the with just the regress block.

Is there such an option? I can’t find it in the help file.

Hi Henry,

The all_runs dataset was created from something, probably pbscale.HEAD. So you could give AP those per-run datasets, instead.

Note that also has -write_3dD_script and _prefix options, for the exact purpose of running a new linear regression model on the same pre-processed data. By adding those options and modifying the -regress options, it is easy to run a new regression model (though the post-regression steps would not be included).

  • rick

The single run preprocessed files are deleted in the cleanup of, so I now longer have them. Obviously the all_runs could be cut up into the runs, and run that way.

As best I understand the -write_3dD_script, it’s just so as to output a 3dDeconvolve script that can be used at a later point, right? It’s not a dry run or anything like that?

I guess that running a 3dDeconvolve script makes the most sense, in that case.

That is right, it just makes a 3dDeconvolve script. You can always add a -concat option to that, maybe even via -regress_opts_3dD, if it does not need to be early in the option list (though it might).

The 3dD script uses a new prefix, so old files are not overwritten (except for 3dDeconvolve.err).

  • rick