CODEWITHSUNDEEP
awkpattern-matchingbioinformaticstext-processing
Carmen Sandoval
Carmen Sandoval

Reputation: 2356

awk: print lines that DO NOT match patterns in a file, looking at a specific column

I have an idFile:

1006006
1006008
1006011
1007002
......

and famFile:

1006 1006001 1006016 1006017 1
1006 1006006 1006016 1006017 1
1006 1006007 0       0       2
1006 1006008 1006007 1006006 2
1006 1006010 1006016 1006017 2
1006 1006011 1006016 1006017 1
1006 1006016 0       0       2
1006 1006017 0       0       1
1007 1007001 1007950 1007015 2
1007 1007002 1007014 1007015 2
......

I need to grep all the lines from famFile where the second column does not match any of the lines in idFile.

This command:

awk 'BEGIN { while(getline <"idFile") id[$0]=1; }
id[$2] ' famFile

returns all the matches:

1006 1006006 1006016 1006017 1
1006 1006008 1006007 1006006 2
1006 1006011 1006016 1006017 1
1007 1007002 1007014 1007015 2
......

But how can I modify the command to get the complement of the matches?

Upvotes: 1

Views: 1653

Answers (1)

James Brown
James Brown

Reputation: 37424

$ awk 'NR==FNR{a[$1];next} !($2 in a)' idFile famFile
1006 1006001 1006016 1006017 1
1006 1006007 0       0       2
1006 1006010 1006016 1006017 2
1006 1006016 0       0       2
1006 1006017 0       0       1
1007 1007001 1007950 1007015 2

Explained:

$ awk '
NR==FNR {                  # process the idFile
    a[$1]                  # hash to a 
    next                   # next id
}
!($2 in a)                 # if the second field id is not in a, output record
' idFile famFile           # mind the file order

Upvotes: 2

Related Questions