retaining text after delimiter in fasta headers using awk - bash

Question

I have what should be a simple problem, but my lack of awk knowledge is holding me back.
I would like to clean up the headers of a fasta file that is in this format:
>HWGG454_Clocus2_Locus3443_allele1
ATTCTACTACTACTCT
>GHW757_clocus37_Locus555662_allele2
CTTCCCTACGATG
>TY45_clocus23_Locus800_allele0
TTCTACTTCATCT
I would like to clean up each header (line starting with ">") to retain only the informative part, which is the second "_Locus*" with or without the allele part.
I thought awk would be the easy way to do this, but I cant quite get it to work.
If I wanted to retain just the first column of text up to the "_" delimiter for the header, and the sequences below, I run this (assuming this toy example is in the file test.fasta):
cat test.fasta | awk -F '_' '{print $1}'
>HWGG454
ATTCTACTACTACTCT
>GHW757
CTTCCCTACGATG
>TY45
TTCTACTTCATCT
But, what I want is to retain just the "Locus*" text, which is after the 3rd delimiter, but, using this code I get this:
cat test.fasta | awk -F '_' '{print $3}'
Locus3443
Locus555662
Locus800
What am I doing wrong here?
thanks.

Answer 1

I understand this to mean that you want to pick the Locus field from the header lines and leave the others unchanged. Then:
awk -F _ '/^>/ { print $3; next } 1' filename
is perhaps the easiest way. This works as follows:
/^>/ { # in lines that begin with >
print $3 # print the third field
next # and go to the next line.
}
1 # print other lines unchanged. Here 1 means true, and the
# default action (unchanged printing) is performed.
The thing to understand here is awk's control flow: awk code consists of conditions with associated actions, and the actions are performed if the condition evaluates to true.
/^>/ is a regex match over the whole record (line by default); it is true if the line begins with > (because ^ matches the beginning), so
/^>/ { print $3; next }
will make awk execute print $3; next in lines that begin with >. The less straightforward part is
1
which prints lines unchanged. We only get here if the first action was not executed (because of the next in it), and this 1 is to be read as a condition that is always true -- nonzero values being true in awk.
Now, if either the condition or the action in an awk statement are omitted, a default is used. The default action is printing the line unchanged, and this takes advantage of it. It would be equally possible to write
1 { print }
or
{ print }
In the latter case, the condition is omitted and the default condition "true" is used. 1 is the shortest variant of this and idiomatic because of it.

Answer 2

$ awk -F_ '{print (/^>/ ? $3 : $0)}' file
Locus3443
ATTCTACTACTACTCT
Locus555662
CTTCCCTACGATG
Locus800
TTCTACTTCATCT

Answer 3

You need a second awk match for the row below. e.g.
cat test.fasta | awk -F _ '/^>/ { print $3"_"$4 } /^[A-Z]/ {print $1}'
Output:
Locus3443_allele1
ATTCTACTACTACTCT
Locus555662_allele2
CTTCCCTACGATG
Locus800_allele0
TTCTACTTCATCT
If you don't want the _allele1 bit remove "_"$4 from the awk script.

Answer 4

You can just do a regex on each line:
$ awk '{ sub(/^.*_L/,"L"); print $0}' /tmp/fasta.txt
Locus3443_allele1
ATTCTACTACTACTCT
Locus555662_allele2
CTTCCCTACGATG
Locus800_allele0
TTCTACTTCATCT