BASH comm command, but for multiple columns

问题

I am looking for something similar to the bash command comm, that I can use to select entries both unique to my 2 files and common to them. Comm worked great when I had just one column per file, eg.

 comm -13 FILE1.txt FILE2.txt > Entries_only_in_file1.txt

But now I have multiple columns of info I wish to keep. I want to select column 2 as the one to filter rows for unique and common entries between my two files. If the entry in column two appears in both files I also want to record the info in columns 3,4,and 5 (if possible, this is not as important). Here is an example of input and output.

FILE1.txt
NM_023928   AACS    2   2   1
NM_182662   AADAT   2   2   1
NM_153698   AAED1   1   5   3
NM_001271   AAGAB   2   2   1


FILE2.txt
NM_153698   AAED1   2   5   3
NM_001271   AAGAB   2   2   1
NM_001605   AARS    3   40  37
NM_212533   ABCA2   3   4   2

Output wanted:

COMMON.txt
NM_153698   AAED1   1   5   3   2   5   3
NM_001271   AAGAB   2   2   1   2   2   1

UNIQUE_TO_1.txt
NM_023928   AACS    2   2   1
NM_182662   AADAT   2   2   1

UNIQUE_TO_2.txt
NM_001605   AARS    3   40  37
NM_212533   ABCA2   3   4   2

I know there has been similar questions before but I can't quite find what I'm looking for. Any ideas greatly appreciated, thank you.

回答1:

join has the following options which are useful for your task:

-j FIELD: join on field FIELD
-o FORMAT: specify output format, as a comma separated list of FILENUM.FIELD.
-v FILENUM: output lines only on FILENUM.

Common to both files:

$ join -j2 -o 1.1,1.2,1.3,1.4,1.5,2.3,2.4,2.5 FILE1.txt FILE2.txt 
NM_153698 AAED1 1 5 3 2 5 3
NM_001271 AAGAB 2 2 1 2 2 1

Unique to FILE1:

$ join -j2 -v1 FILE1.txt FILE2.txt 
AACS NM_023928 2 2 1
AADAT NM_182662 2 2 1

Unique to FILE2:

$ join -j2 -v2 FILE1.txt FILE2.txt 
AARS NM_001605 3 40 37
ABCA2 NM_212533 3 4 2

回答2:

You can archieve that with gnu awk, here is a script:

script.awk

function unique(filename, line) {
    split( line , tmp, FS)
    print tmp[1], tmpp[2], tmp[3], tmp[4], tmp[5] >> filename
}

NR == FNR { # in case we are reading the first file: store line under key
        file1[ $2 ] = $0
        next
    }

    {
        if( $2 in file1 ) { # key from file2 was in also in file1:
            split( file1[ $2 ], tmp, FS)
            print $1, $2, tmp[3], tmp[4], tmp[5], $3, $4, $5 >> "COMMON.txt"
   # remove common key, thus we can later find unique keys from file1
            delete file1[ $2 ] 
        }
        else { # unique key from file2 
            unique("UNIQUE_TO_2.txt", $0)
        }
    }

END {
  # remaining keys are unique in file1
        for( k in file1 ) {
            unique("UNIQUE_TO_1.txt", file1[ k ])
        }
    }

Use it like this:

# erase the output files if present
rm -f COMMON.txt UNIQUE_TO_1.txt UNIQUE_TO_2.txt
# run script, create the file
awk -f script.awk FILE1.txt FILE2.txt
# output the files
for f in COMMON.txt UNIQUE_TO_1.txt UNIQUE_TO_2.txt; do echo "$f"; cat "$f"; done

The printf ... >> filename appends the text to filename. This requires the rm of the output files when running the script a second time.

来源：https://stackoverflow.com/questions/37170326/bash-comm-command-but-for-multiple-columns

标签

bash

unique

comm