[SOLVED] grep semi-variable pattern

Drosera_capensis · 09-08-2021, 03:17 AM

Hello everyone,

I have used grep to process data from this input:

Code:

Scaffolds_16_pilon      ACmerged_contig_2215    5_1009.75816    11_1010 11_1011

To that output:

Code:

Scaffolds_16_pilon      ACmerged_contig_2215    1    5   
Scaffolds_16_pilon      ACmerged_contig_2215    2    11

Part of my script is based on selection the correct first number before the underscore in column three and four.
For this I have used the regexp '$f'_.* to select a specific $f followed by underscore, then a variable pattern.

Code:

grep --only-matching '\<'$f'_.*\>' |  wc --word

Unfortunately, when $f=5 the script select the three last column and not just the first one. And I don't really understand why. Any idea anyone?

Drosera_capensis · 09-08-2021, 03:24 AM

A colleague just gave me this answer:

Quote:

grep -o ''$f'_' file.txt

shruggy · 09-08-2021, 04:03 AM

Quote:

Originally Posted by Drosera_capensis

Unfortunately, when $f=5 the script select the three last column and not just the first one. And I don't really understand why.

Because the * quantifier is greedy. _\S* would select to the end of field.

Turbocapitalist · 09-08-2021, 04:10 AM

Quote:

Originally Posted by Drosera_capensis

Unfortunately, when $f=5 the script select the three last column and not just the first one. And I don't really understand why. Any idea anyone?

If the data is it tab-delimited columns, then you might also consider AWK. It has lesser pattern matching capabilities than grep but has other advantages. Perl is another option if advanced patterns are needed.

Code:

export f=5; 

awk -v f=$f '$3~f { print $1,$2 }' FS="\t" OFS="\t" ./data.tab

perl -0x08 -a -e 'chomp; if($F[2] =~ /$ENV{"f"}/){ printf("%s\t%s\n", $F[0],$F[1]); }' ./data.tab

pan64 · 09-08-2021, 04:48 AM

Code:

grep -o -P "$f"'_\S*'