0

I have a small snippet of a file I'm working with:

ENSDARG00000032737    ENSDARP00000120731
ENSDARG00000032737    ENSDARP00000049290
ENSDARG00000061051    ENSDARP00000081062
ENSDARG00000061051     
ENSDARG00000061051    ENSDARP00000129708

I only want to print the first instance of each unique value in the first column and the corresponding value in the second column, so my desired output would be:

ENSDARG00000032737    ENSDARP00000120731
ENSDARG00000061051    ENSDARP00000081062

Is there a simple way to accomplish this with awk or uniq or something similar?

Any help would be appreciated.

4
1

This idiomatic solution will work robustly using any awk in any shell on every UNIX box:

$ awk '!seen[$1]++' file
ENSDARG00000032737    ENSDARP00000120731
ENSDARG00000061051    ENSDARP00000081062
1
  • 2
    +1 Very concise and "idiomatic", as well as strictly equivalent to StevenPenny's answer (not at all meant as a criticism), but nevertheless a little abstruse at first sight for those who do not speak "awk" fluently. ;-)) – Cbhihe Apr 4 '20 at 18:08
2

POSIX AWK:

m1[$1] == 0 {
   m1[$1] = 1
   print
}

For each line:

  1. see if first column exists in the "database"
  2. if not, add to "database" and print entire line
2
$ sort -s -k1,1 -u file
ENSDARG00000032737    ENSDARP00000120731
ENSDARG00000061051    ENSDARP00000081062

This sorts the file based on the first column only. While doing so, it ignores lines whose first column has already been seen.

Most implementations of sort has a non-standard -s option (used in the command above) that guarantees that it will be using a "stable" sorting algorithm. A stable sorting algorithm does not change the ordering of entries that have identical keys (first column in your case).


Note however that the longer transcript (which both Ensembl and Havana agrees 100% on) for the ENSDARG00000032737 gene is ENSDART00000049291, which codes for ENSDARP00000049290, not ENSDARP00000120731. But that's not really my business.

1
  • 1
    Thanks! I appreciate the feedback on the bioinformatics too, but your solution should work just fine for my purposes. – gpreising Apr 3 '20 at 22:35
0

Best solutions already provided just posting my try

for i in `awk '{if(!seen[$1]++)print $1}' filename`; do sed -n '/'$i'/{p;q}' filename; done

output

ENSDARG00000032737    ENSDARP00000120731
ENSDARG00000061051    ENSDARP00000081062

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.