Hi,
I've been trying to remove the letters "chr" from fasta headings of hg19 ref. genome. I tried a following sed script;
sed 's/[chr,]//g'
I am just wondering if there are any other sed or awk scripts to the same.
Thanks.
You're streaming the contents of a FASTA file and you would like to edit this stream. sed is your best bet. If you'd like to edit the file in-place and not create a new file with the output of sed, you'll want to try sed -i. If you're on a Mac and would like to use GNU-sed specific options, you might wanna use the gsed utility.
Quick question: Why is the search RegEx [chr,] and not just chr?
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
Try more. Here is a link that can help you http://www.brunolinux.com/02-The_Terminal/Find_and%20Replace_with_Sed.html
Ashutosh, unless you use -i, all sed commands write to stdout and do not affect the original file.
Yes, you are right. It was a warning for newbies who may copy and use a wrong command and accidentally modifying their original file.
I agree, but IMHO, we must advice them to check for the -i option rather than saying that sed overwrites files. Anyway, moot point as OP looks pretty at home with UNIX (grep at least)
Why other scripts, does this not work?
I am in a learning process. By looking at a wide rage of sed and awk scripts having same functions on the web, I am just curious to see what alternatives are. Plus I am not yet sure this script fully works.
I usually stick to sed for file content editing. My motto is to use a tool that we all know works. We could use Perl or Python or awk for this, but I'm comfortable with the simplicity and integration that sed gives me.
I like your approach though - makes for quite a bit of learning!
Another good resource for find, grep, sed and awk: http://wilsonericn.wordpress.com/2011/08/25/find-grep-sed-and-awk/
There may be something wrong with this script. A size of output.fa is around 2.86GB whereas the original input.fa is around 3.16GB. I don't think eliminating the three characters from headers would result in the loss of nearly 300kb ?
This script did delete 'chr' from the header showing the result of
and
Am I missing something here?
That sure sounds a bit unexpected. There are only 25 headers and 75 characters should not amount to a 300 MB difference.
Is there a reason for the
[chr,]usage though? Why not justchr?Also, maybe a difference in the
wc -cvalues before and after the script would make it clear of the number of characters being deleted is >75. This is an expensive process though.Hi,
For
[chr,], I was just following some examples on the net. I thought[]was necessary to specify certain letters/characters in the word (in this case 'chr' at begging of the word) but I was wrong. I tried just'chr'and really worked. Overall, the file size remains similar. It will be long way to master a usage of sed.By 'file size remains similar', you mean that using
[chr,]orchrdoesn't make a difference, right?They did make the significant difference. As in my previous post, the script with
[chr,]chewed up around 300kb whereas the one with'chr'did not change the file size before and after.Ah, of course, the expression
[chr,]deleted allCs,Hs,Rs and,s. This effectively eliminated 1/4th of the genome sequence. How could I not have seen that![chr,]translates to 'c' or 'h' or 'r' or ','.'chr'translates to the exact string 'chr'.