If you have a bash shell, this will work...
Code:
cat infile | cut -c101-135,136-160,350-420 | tr '[:upper:]' '[:lower:]' | sed 's/ \s\+/,/g' > outfile
uniq infile outfile
EDIT: This was tested using the "al_1st-al.exe" file from the site you linked to. Only 8 dupe lines are dropped.