Look for the -blip_forward_dset and -blip_reverse_dset options in afni_proc.py. In this case, you'll need to do an analysis of the L-R data with the reverse blip (R-L) and then an afni_proc separately for the R-L data with reverse blip (L-R). The blip portion of the processing actually happens fairly early on, so you could modify the afni_proc output script to run the data through the first few steps of each run and then pickup the analysis with both corrected runs afterwards.
I've not tried the blip correction on L-R/R-L data personally, but I am not coming up with reasons why it wouldn't work. Post followups if you have hiccups!
-Peter
