It's not too surprising the oblique EPIs do not overlay the anatomical datasets. You can try 3dWarp by itself to deoblique the EPI data to match the anatomical datasets or oblique the anatomical datasets to match the EPI dataset. Though this doesn't do alignment, it will provide a rough idea of whether they are in the same position after alignment. See the following link on generally how to handle oblique data:
[
afni.nimh.nih.gov]
[
afni.nimh.nih.gov]
In some cases, the oblique or cardinal coordinate systems may not be well defined in the original DICOM files, so the obliquing/deobliquing may not work properly. In that case, the best solution is to use the "-giant_move" and "-deoblique off" with align_epi_anat.py and let the alignment move the datasets around.