Remove columns using shell commands

With GNU awk 4.0 or later:

gawk 'flag { split($0, f, " ", d); for(i = 1; i <= 5; ++i) printf("%s%s", d[i - 1], f[i]); print ""; next } /@<TRIPOS>ATOM/ { flag = 1 } 1' filename

Most of this is to keep the formatting intact; if the formatting does not matter, then

awk 'flag { NF = 5 } /@<TRIPOS>ATOM/ { flag = 1 } 1' filename

is a simpler way that works with older gawk and mawk as well. To make this work with BSD awk,

awk 'flag { NF = 5; $1 = $1 } /@<TRIPOS>ATOM/ { flag = 1 } 1' filename

is necessary ($1 = $1 just to force the rebuilding of the line). Thanks to @tripleee for commenting on this.

The second piece of code simply adjusts the number of fields, causing the line to be rebuilt with less of them. The first does a bit more:

flag {                              # if we're already processing lines
  split($0, f, " ", d)              # split line into array f, save delimiters
                                    # into array d

  for(i = 1; i <= 5; ++i) {         # print the first five fields separated
    printf("%s%s", d[i - 1], f[i])  # by the saved delimiters
  }
  print ""                          # add newline
  next                              # that is all.
}
                                    # if we're not processing lines yet
/@<TRIPOS>ATOM/ { flag = 1 }        # check if we should, and if so set flag
1                                   # then print line unchanged.

Addendum: Another way that also preserves the formatting is to use sed:

sed '1,/@<TRIPOS>ATOM/ ! { s/\b[[:space:]]/\n/5; s/\n.*//; }' filename

That is:

1,/@<TRIPOS>ATOM/ ! {     # For those lines that are not in the range from
                          # the beginning to the first line containing
                          # @<TRIPOS>ATOM

  s/\b[[:space:]]/\n/5    # place a newline after the fifth column
  s/\n.*//                # then remove the newline and everything after it
}

This should work with both GNU sed and BSD sed. Since \b is not part of POSIX basic regexes, though, more esoteric seds may require a slight change:

sed '1,/@<TRIPOS>ATOM/ ! { s/\([^[:space:]]\)[[:space:]]/\1\n/5; s/\n.*//; }' filename

This works essentially the same way but uses a different regex to recognize the end of columns.

The following should work:

sed -n '/@<TRIPOS>ATOM/,$p' filename | tail -n +2 | tr -s " " | cut -d" " -f1-5

Work as follows:

1. Print only the lines after the @<TRIPOS>ATOM:

   sed -n '/@<TRIPOS>ATOM/,$p' filename
   

2. Omit the first line (which contains @<TRIPOS>ATOM and you don't want that):

   tail -n +2
   

3. Squeeze the extra spaces between columns:

   tr -s " "
   

4. cut the columns using space as the delimiter, grab the fields you need:

   cut -d" " -f1-5