How to replace strings in the last n-lines of multiple files in linux

How can I replace strings in lets say 'last 10 lines' of multiple files?

I have about 100 files with the same extension '.txt' and I would like to replace string 'GLN' to 'LOO' in the last 10 lines of each file. How do I do this? I know how to do it for one file but not for several files. When I use this command ;

for i in `head -3 *.txt  | awk '{print $4}'`
     do
                   sed -i 's/GLN/LOO/g' *.txt 
     done

It replaces GLN everywhere it appears in the files and not just the last 10 lines. Please, what am I doing wrong?

It's not clear what head -3 *.txt | awk '{print $4}' returns here since you haven't shown us a sample of your file(s), however sed -i 's/GLN/LOO/g' *.txt will replace all of the instances of GLN in all the lines in all the files matching *.txt, provided the loop executes at least once.

AFAIK there's no direct way to address the last n lines of a file in sed - so to use that, you would need to calculate the offset externally, using wc -l and shell arithmetic for example:

for f in *.txt; do 
  start=$(( $(wc -l <"$f") - 9 ))
  sed "$start"',$s/GLN/LOO/g' "$f"
done

(I removed the -i so output goes to the terminal, for testing). You could use something like awk 'END{print NR-9}' "$f" in place of $(( $(wc -l <"$f") - 9 )) to get the start offset if your intention was to use awk.

Alternatively you could use tac to invert the file and make replacements in the first n lines, then tac the result - although that makes in-place replacement complicated.

In this situation I would probably reach for ed or ex which do support numerical address offsets ex. testing with output to the terminal:

for f in *.txt; do 
  printf '%s\n' '$-9,$s/GLN/LOO/g' ',p' | ed -s "$f"
done

Once you are happy that it's doing the right thing, change ,p to wq to write the result to the file (equivalent of sed's -i).

I would use tac and awk functions for this job as follows:

tac file1.txt | awk 'NR<11 {gsub("GLN","LOO")};{print}' | tac