A couple of things that you can consider:
Reduce the number of time frames you analyze (ie, increase nskip)
Do the loop over amino acids inside the trajectory-reading loop (ie, read a frame and then analyze the interactions for all 12 aa before reading the next frame)
CONS FIX sele .not. segid prt end after you have read one frame
then CONS FIX sele none end before you read the next frame:
label tloop
cons fix sele none end
traj read
cons fix sele .not. segid prt end
intere ...
write title unit 101
intere ...
write title unit 102
.
.
intere ...
write title unit 112
goto tloop
And you can of course do all 96 trajectories at the same time if you have enough cores.
I would add ".and. segid PRT" to the first intere-selection, just in case there are solvent molecules with the same resid as one of the amino acids.
Last edited by lennart; 04/25/19 12:54 PM. Reason: Added CONS FIX