awk regular expression

Shell Programming and Scripting
1

Hello,

I have big files which I wanna filter them based on first column.

first column should be one of these strings: chr2L || chr2R || chr3L || chr3R || chr4 || chrX
and something like chr2Lh or chrY or chrM3L is not accepted.

I used the following command:

 awk '{ if ($1=="chr2L" || $1=="chr2R" || $1=="chr3L" || $1=="chr3R" || $1== "chr4" || $1=="chrX") print $0 }' input | sort -k1,1 -k2,2n > output

Is there easier way to do it? or is it true anyway?!!
And BTW, I wanna sort based on two column which the first is string and second is number and you see what I used in the code above. Please let me know if it's true also.

Thanks in advance.

2

You could do:

awk '$1 ~ /^chr[2-4]*[LRX]*$/' input | sort -k1,1 -k2,2n

EDIT: But If you want to match strictly only chr2L || chr2R || chr3L || chr3R || chr4 || chrX , then above regexp is wrong.

It will match other combinations as well. I think what you did is correct.

1 Like
3

I was about to respond to this post to say exactly what you added in your edit. I agree completely.

Shorter isn't always better. That said, the original version can be simplified a little bit:

 awk '$1=="chr2L" || $1=="chr2R" || $1=="chr3L" || $1=="chr3R" || $1=="chr4" || $1=="chrX"' input | sort -k1,1 -k2,2n > output

Regards,
Alister

1 Like
4

thank you Yoda and alister.

As you said shorter is not always better! :slight_smile: I was not sure if what I used does the matching strictly or not. Now I'm
once again thanks to both of you.

Cheers!

5

Not sure if this will work with all awk version, but you might want to try

awk '$1 ~ /^(chr2L|chr2R|chr3L|chr3R|chr4|chrX)$/' input | sort -k1,1 -k2,2n

It's an or ed regex constant fitting exactly within the bounds of $1.

6

Factoring out the common string chr :

awk '$1 ~ /^chr(2L|2R|3L|3R|4|X)$/' input | sort -k1,1 -k2,2n

Try to avoid regex's if you can, to improve efficiency. Also, if you happen to know, to a reasonable extent, the frequency of occurrence of those strings, you could use simple string comparisons (with the proper order of comparisons) with the short-circuit logical OR ( || ) operator.

1 Like