Cleaning transcripts before running requires these steps:
1.  Run status.sh and pull.sh to update the local database

2.  CLAN: In order to make things run more smoothly, it is best to remove the %mor and %gra which are easy to add back later.  
trim -t%mor -t%gra +re *.cha +1
longtier *.cha +1

3. CLAN: expand the [x N] notations folder by folder with this CLAN command:
kwal -d90 +d +f +t@ +t% *.cha +1 +re

4. Fix old uses of & to &~ using temp in CLAN

5.  For English only, remove pluses from compounds in the Terminal using
noplus.pl *.cha
find /path/to/chat/dir -name '*.cha' -print0 | xargs -0 noplus.pl 

6.  CLAN: run rep-comp.cut and rep-join.cut with CHSTRING. They are in /fixes
chstring +crep-comp.cut *.cha +1
chstring +crep-join.cut *.cha +1
chstring +crep-under.cut *.cha +1

7. CLAN: run MOR +xb *.cha to make sure all compounds were recognized
add any missing ones to the lexicon

8. BBEdit: to remove repetition marks next to fillers and fragments (both & and &+)
Filler1: \&\+([a-z]+) \[\/\] \&\+([a-z]+) Replace: \&\+\1 \&\+\1  (with grep)
Filler2:  \&([a-z]+) \[\/\] \&([a-z]+)  Replace: \&\1 \&\1  (with grep)
Find: &-um [/] &-um Replace: &-um &-um  (without grep)
Find: &-uh [/] &-uh Replace: &-uh &-uh (without grep)

9.  run Chatter and fix errors

10.  remove all *.ulx files

11:  If there is no audio, you are finished, if there is audio, continue with batchalign
(assumes miniconda is installed and batchalign is downloaded)
conda activate batchalign
batchalign ~/mfa_data/input ~/mfa_data/output --prealigned  (transcript has bullets)
or
if transcripts have audio but no bullets, then use F5 to hand align and then run above command