Replacing Dots In File

Question

[Patrick]

Hi to all

I have a file like this

chr1    9992    10466    chr1    10240    10349    .    0    +    6.679490    4.52317    -1
chr1    11026    11535    .    .    .    .    .    .    .    .    .
chr1    11560    14749    .    .    .    .    .    .    .    .    .

I would like to replace dots by zeros without replacing float values that contains . any trick to do this with perl/sed or awk ?

Many thanks

Pat

[toni]

Why not replacing occurrences of "t." by "t0" ?

[Jonathan Cline Ieee]

If the input contains malformatted float values, then numbers might begin with dot... i.e. " .123 "

Answer 1

[toni]

sed 's/\t\./\t0/g' file.txt > output.txt

output.txt is :

chr1    9992    10466   chr1    10240   10349   0   0   +   6.679490    4.52317 -1
chr1    11026   11535   0   0   0   0   0   0   0   0   0
chr1    11560   14749   0   0   0   0   0   0   0   0   0

Answer 2

[Malachi Griffith]

With Perl

 perl -ne '$_ =~ s/\t\./\t0/g; print $_' file.txt > output.txt

With Awk

awk '{ gsub("\t\\.", "\t0"); print }' file.txt  > output.txt

[Chris Penkett]

Or to replace in-place with perl and minimal code: perl -pi~ -e 's/t./t0/g' file.txt

Answer 3

[Jonathan Cline Ieee]

The other answers will change the field separator rather than preserving it. So if files have tabs vs. spaces, this could break the file. Better is to preserve the separator in the replacement text. Although that's a more complex pattern (Perl):

s/(\s)\.(\s)/${1}0${2}/g;

or

$ perl -pe 's/(\s)\.(\s)/${1}0${2}/gx;' <dat

Likely you hit the problem of using 102 which could be interpreted as backreferencing the 10th match (although Perl isn't supposed to behave that way..).

You asked for perl/sed/awk, however it's also very easy in Vim. Vim is handy for this because you can experiment with bunch of regex's and then hit Undo until you finally get what you want.

• load the file

• Use the regex command as follows.

  :%s/ . / 0 /g

or to preserve the separator,

:%s/\(\s\)\.\(\s\)/\10\2/g