Keeping the characters between two specific characters in fasta headers and removing the rest

Question

I am trying to keep the characters in my fasta headers between the first and the second colon. I've failed trying with sed and awk since my coding skills are very limited.

In:

>TRINITY_DN10034_c0_g1:TRINITY_DN10034_c0_g1_i1:g.1561:m.1561
CCAGCGCCACGGAGGGCGAAGGCGAAGGCGGTGGTTTTGAACGATGAGGTGCCGCCGGTG
>TRINITY_DN10322_c0_g1:TRINITY_DN10322_c0_g1_i2:g.1424:m.1424
GATCCGCCTTCTATAACTGGAAAAGAAGAGGATGAGGTGCCGCCGATGAGTCTACCTTCT

Expected:

>TRINITY_DN10034_c0_g1_i1
CCAGCGCCACGGAGGGCGAAGGCGAAGGCGGTGGTTTTGAACGATGAGGTGCCGCCGGTG
>TRINITY_DN10322_c0_g1_i2
GATCCGCCTTCTATAACTGGAAAAGAAGAGGATGAGGTGCCGCCGATGAGTCTACCTTCT

Any help?

Answer 1

With awk:

awk -F: '{ if (NF==1) { print } else { print ">"$2 } }' file

or a little more compressed

awk -F: '{ print (NF==1) ? $1 : ">"$2 }' file

Use field separator : and print the first field if the line has only one field and print > and the second field otherwise.

With sed:

sed 's/^>[^:]\+:\([^:]\+\).*/>\1/' file

Answer 2

With standard sed:

$ sed '/^>/{ s/[^:]*:/>/; s/:.*//; }' file.fa
>TRINITY_DN10034_c0_g1_i1
CCAGCGCCACGGAGGGCGAAGGCGAAGGCGGTGGTTTTGAACGATGAGGTGCCGCCGGTG
>TRINITY_DN10322_c0_g1_i2
GATCCGCCTTCTATAACTGGAAAAGAAGAGGATGAGGTGCCGCCGATGAGTCTACCTTCT

This applies the two substitutions s/[^:]*:/>/ and s/:.*// to all lines in file.fa that starts with a > (the fasta header lines).

The first substitution would replace the start of the line, up to and including the first : character, with >.

The second substitution would remove everything from the first : in the resulting data (the second : in the original line) to the end of the line.

This would leave the second :-delimited field, prefixed by a >.

---

With standard awk:

$ awk -F ':' '/^>/ { $0 = ">" $2 } { print }' file.fa
>TRINITY_DN10034_c0_g1_i1
CCAGCGCCACGGAGGGCGAAGGCGAAGGCGGTGGTTTTGAACGATGAGGTGCCGCCGGTG
>TRINITY_DN10322_c0_g1_i2
GATCCGCCTTCTATAACTGGAAAAGAAGAGGATGAGGTGCCGCCGATGAGTCTACCTTCT

This would use : as a field delimiter, and when coming across a line that starts with >, it would replace the complete line with > followed by the second :-delimited field. All lines, whether modified or not, would then be printed.

Answer 3

A propposed solution:

With GNU sed

cut -f2 -d':' file | sed '1~2s/^/>/'

---

Output:

>TRINITY_DN10034_c0_g1_i1
CCAGCGCCACGGAGGGCGAAGGCGAAGGCGGTGGTTTTGAACGATGAGGTGCCGCCGGTG
>TRINITY_DN10322_c0_g1_i2
GATCCGCCTTCTATAACTGGAAAAGAAGAGGATGAGGTGCCGCCGATGAGTCTACCTTCT

Answer 4

$ awk -F':' '{$0 = (NF>1 ? ">"$2 : $0)} 1' file
>TRINITY_DN10034_c0_g1_i1
CCAGCGCCACGGAGGGCGAAGGCGAAGGCGGTGGTTTTGAACGATGAGGTGCCGCCGGTG
>TRINITY_DN10322_c0_g1_i2
GATCCGCCTTCTATAACTGGAAAAGAAGAGGATGAGGTGCCGCCGATGAGTCTACCTTCT

$ awk -F':' '{sub(/>.*/,">"$2)}1' file
>TRINITY_DN10034_c0_g1_i1
CCAGCGCCACGGAGGGCGAAGGCGAAGGCGGTGGTTTTGAACGATGAGGTGCCGCCGGTG
>TRINITY_DN10322_c0_g1_i2
GATCCGCCTTCTATAACTGGAAAAGAAGAGGATGAGGTGCCGCCGATGAGTCTACCTTCT

Answer 5

There are a multitude of ways to handle this pbm, some of them being:

$ cut -d: -f2 | sed -e '$!N;s/^/>/' file

$ awk -F: '/^>/{$0 = ">" $2}1' file

$ perl -F: -pale ' s/^>\K.*/$F[1]/' file

$ sed -e '
   /\n/{P;d;}
   /^>/y/:/\n/
   s/\n/&>/;//D
' file

Results

>TRINITY_DN10034_c0_g1_i1 
CCAGCGCCACGGAGGGCGAAGGCGAAGGCGGTGGTTTTGAACGATGAGGTGCCGCCGGTG
>TRINITY_DN10322_c0_g1_i2 
GATCCGCCTTCTATAACTGGAAAAGAAGAGGATGAGGTGCCGCCGATGAGTCTACCTTCT