The way that AFNI was intended to be used is that all the datasets from a single scanning session would be stored in a separate directory (a "session directory"). These datasets are presumed to be aligned with one another. When you mark and transform one of them, it is assigned to be the anatomy parent of the other datasets in that directory.

Note This Carefully:
If you stored datasets from multiple sessions in the same directory, AFNI won't know which marked/transformed dataset is the correct anatomy parent to use for which unmarked datasets. For example, if you have scanned two people, Fred and Ethel, you might have the following datasets:

  ethel.anat+orig    ethel.anat+acpc    ethel.anat+tlrc
  ethel.func+orig
  fred.anat+orig     fred.anat+acpc     fred.anat+tlrc
  fred.func+orig

You can fix this problem by assigning the correct anatomy parent datasets. This can be done at dataset creation time with to3d, or it can be done later using the -apar option of program 3drefit (this option was added 14 Oct 1999). In the case of Fred and Ethel, the correct command lines would be

  3drefit -apar fred.anat+orig fred.func+orig
  3drefit -apar ethel.anat+orig ethel.func+orig

Note that if you use program adwarp to compute follower datasets, then you are required to give the anatomy parent on the command line, and so the problem discussed here shouldn't arise (assuming you use the correct dataset for the parent!).

I have heard that some people think that when AFNI is used to write out a functional dataset in Talairach coordinates, the current Anat dataset is used to master the coordinate transformation. This assumption is incorrect! The anatomy parent of the functional dataset is used; if one was not assigned by the user (via to3d or 3drefit), then the first anatomical dataset with a transformation in each directory will be chosen as the anatomy parent for all "orphan" datasets in that directory. This means that when combining multiple scanning sessions, even of the same subject, in the same directory, you must be careful to set the anatomy parents correctly. Otherwise, you will get incorrect transformations to Talairach coordinates applied to some of your functional datasets.

As of 03 Dec 1999, AFNI now prints a warning when a dataset is "forcibly adopted" to have an anat parent. This warning is issued when (a) the dataset does not have an anat parent encoded into its header file, and (b) there is more than one anatomical dataset with Talairach markers in the same directory.

As of 15 Dec 1999, most AFNI programs that create a dataset from another dataset (e.g., the "3d" codes) will make the anat parent ID of the output dataset be the same as the anat parent ID of input dataset. If this relationship needs to be altered, this can be done using 3drefit.

When a dataset has an anat parent ID encoded in its header, the interactive AFNI program will attempt to match the dataset to its anat parent at startup. If it cannot make the match, a warning message will be printed. This is the only program that actually uses the anatomy parent encoded in the dataset header (at this time), since it is the only program that reads in a whole library of datasets. The other program that uses the anatomy parent idea is adwarp, but that program requires input of the chosen anatomy parent on the command line - it will ignore the anatomy parent ID in dataset headers.

[Answer added 14 Oct 1999; edited 16 Dec 1999]