renaming all fasta headers in a file

Question

[branokdrung]

Hi everyone,

I'm encountering a problem with too long fasta headers. They get truncated at the 20th position by a program (TargetP) I'm using.

Example:

>ConsensusfromContig10000-snap_masked-ConsensusfromContig10000-abinit-gene-0.1-mRNA-1:cds:3144/1451-1467:0:+
MKKSGDIDEIWKSMQEDARPKPRLPPLPAAAPPAPAPPAPAPKAAAAQPAAASSSNAMVAVNGGASRAFDYSNANALQRDINSLGDEALGTRKRAAERLEAVIVGAEGEAAEATVRALTGDLFKPLLKRFADPGEK

What remains are thousands of entries named "ConsensusfromContig1".

Is there any software or any script I can use to rename the headers in a way that they are 20 characters long and still able to get identified? I have only found scripts for truncating too long headers so far. The desired naming for the example would be something like 10000|3144/1451-1467:0 .

I would be grateful for any help provided.

Answer 1

[Anima Mundi]

In Python:

for line in open('input.fa'):
    if '>' in line:
        r_line = line[::-1]
        r_header = r_line[1:19]
        print '>' + r_header[::-1]
    else:
        print line,

Answer 2

[iraun]

If you have always the same format of header line, I mean, always Contig word and cds word, you can use this awk command:

awk '{if($1 ~ /^>/){split($1,a,"-"); split(a[1],b,"Contig");split($1,c,"cds:"); print ">"b[2]"|"c[2]}else{print}}' file

[branokdrung]

Thanks a lot! Never imagined it could be done so easy. I used your command in the following way:

awk '{if($1 ~ /^>/){split($1,a,"-"); split(a[1],b,"Contig");split($1,c,"cds:"); print ">"b[2]"|"c[2]}else{print}}' Cyanophora_paradoxa_MAKER_gene_predictions-022111-aa.fasta >> Cyanophora_paradoxa_MAKER_gene_predictions-022111-aa-newHeaders.fasta

It worked like a charm. Big thanks again!