Keeping first instance of duplicates when column values are interchangeable

Question

I have previously asked a question about keeping the first instance of duplicates and the solution provided has been working perfectly until now (see Keeping first instance of duplicates).

However, I now have the scenario where the values in Columns C and D could be the same but in different orders and I only want to keep one (either order will do).

Example input:

A B C D E F G
1 2 T TACA 3 2 Q
9 3 A C 9 3 P
8 3 I R 8 2 Q
9 3 C A 9 3 P
4 8 C T 7 4 P
9 3 T G 9 3 P

Desired output:

A B C D E F G
1 2 T TACA 3 2 Q
9 3 A C 9 3 P
8 3 I R 8 2 Q
4 8 C T 7 4 P
9 3 T G 9 3 P

Using: sort -k3,4 -k5,5r -k1,1r file | sort -k1,1 -k3,4 -u | sort -k1,1r (or similar) keeps both the 9 3 A C 9 3 P and 9 3 C A 9 3 P versions but I only want to keep one of these. The added complication is that for values for Column A could have multiple acceptable Column C and D values (such as row 2 9 3 A C 9 3 P and row 5 9 3 T G 9 3 P in the Desired Output example above) so cannot search for duplicates based on Column A alone.

Thanks!

Answer 1

awk '{
    key1 = $1 FS $2 FS $5 FS $5 FS $7
    if ( ((key1 SUBSEP $3 FS $4) in seen) || ((key1 SUBSEP $4 FS $3) in seen) )
        next
    seen[key1, $3 FS $4] = 1
    print
}'

A B C D E F G
1 2 T TACA 3 2 Q
9 3 A C 9 3 P
8 3 I R 8 2 Q
4 8 C T 7 4 P
9 3 T G 9 3 P