Linux sort -u does not work - sorting

Question

I tried to sort a big file with more than 10000 data. I want to find the unique species (like Fe, La, etc.) in the file (17x200.o_neighbors.raw.dat) Ideally, I should get the result like below (see the 4th column)
FRAME 0 9194 Fe 6330SI
FRAME 11 9194 La 12858H 6330SI
However, I got results like this
FRAME 0 9194 Fe 6330SI
FRAME 11 9194 La 12858H 6330SI
FRAME 19 9194 La 13537H 6330SI
There are two "La" species. How can I get the duplicated one removed.
Here is my command
grep FRAME 17x200.o_neighbors.raw.dat | grep 9194 |sort -k 2 -n |sort -k 4 -u
the first sort -k 2 -n is to get timeseries order
the second sort -k 4 -u is to get unique species data
Any suggestion would be appreciated.

Answer 1

Per the sort manpage, the -k flag works like so:
-k, --key=POS1[,POS2]
start a key at POS1 (origin 1), end it at POS2 (default end of line)
So -k 4 defines a key from position 4 to the end of the line; so in your example, its values are { Fe 6330SI, La 12858H 6330SI, La 13537H 6330SI }, which are all distinct.
To fix this, you need to define a key from position 4 to position 4:
... | sort -k 4,4 -u

Answer 2

Using awk:
$ awk '($4 in a==0) { # if $4 not hashed yet, ...
a[$4]=$0 # hash it to a
}
END { # after all record have been processed
for(i in a) # iterate all hashed records
print a[i] # output
} ' file
FRAME 0 9194 Fe 6330SI
FRAME 11 9194 La 12858H 6330SI
Now you can sort that output.

Answer 3

I test it. I use command below to filter the duplicated one. But I do not understand why.
grep FRAME 17x200.o_neighbors.raw.dat | grep 9194 |sort -k 2 -n |sort -k 3,4 -u
Any explanation would be appreciated.