Text to fasta and other delights of the shell

One thing I’ve learned in my current job is that some familiarity with Linux tools for processing text files: awk, sed, grep, head/tail, cut/paste and so on, often provides a speedier solution than writing a script in (insert scripting language of choice here). I know this stuff is trivial to shell gurus, but I still get a little buzz out of it. A couple of real-life examples.

1. Delimited text to fasta

Your collaborator has developed “peptides on a chip” and sends you a text file describing said chip where column 1 = spot number, column 2 = peptide sequence:

1	ELDQGSLATSF
2	VDLAATPTDVR
3	TTNEEYLDLSQ
4	SSLDVYDGRFL
5	ALVHSYMTGRR

Your analysis requires that the peptide sequences be in fasta format. Easy, using awk:

awk '{print ">peptide" $1 "\n" $2}' infile > outfile

Result:

>peptide1
ELDQGSLATSF
>peptide2
VDLAATPTDVR
>peptide3
TTNEEYLDLSQ
>peptide4
SSLDVYDGRFL
>peptide5
ALVHSYMTGRR

2. Delimited text to sorted files named by field

Output from a program to predict phosphorylation sites for protein kinases contains 6 columns – kinase name, domain number, peptide name, residue position, sequence and score, like so:

kin2	1	pep2	7	RASTFAG	89.82
kin2	1	pep1	6	RRTSLGT	82.69
kin1	1	pep1	6	RAVSMDN	82.22
kin1	1	pep2	7	RASSDGE	80.06
kin2	1	pep1	6	RHSSYPA	80.09
kin1	1	pep2	7	RTTSFAE	82.27
kin2	1	pep1	6	RAASMDS	84.43
kin1	1	pep1	6	RRLTFRK	80.51
kin2	1	pep2	7	RASSADD	84.74
kin1	1	pep2	7	RRASVAA	84.59

You want to sort the output into separate files in two ways:

• By kinase: show the best substrates for a kinase
• By substrate: show the best kinases for a substrate

Awk to the rescue again, with sort along for the ride. Case 1 – best substrates for a kinase:

awk '{file = $1; command = "sort -nrk6 >>"; print $0 | command file}' infile

Result: two files named kin1 and kin2, showing substrates for the respective kinases, sorted by score. File kin1 looks like this:

kin1	1	pep2	7	RRASVAA	84.59
kin1	1	pep2	7	RTTSFAE	82.27
kin1	1	pep1	6	RAVSMDN	82.22
kin1	1	pep1	6	RRLTFRK	80.51
kin1	1	pep2	7	RASSDGE	80.06

Case 2 – best kinases for a substrate, is as simple as substituting $3 (substrate name) as the output file:

awk '{file = $3; command = "sort -nrk6 >>"; print $0 | command file}' infile

Result: two files named pep1 and pep2, showing kinases for the respective substrates, sorted as before. Here’s the contents of pep1:

kin2	1	pep1	6	RAASMDS	84.43
kin2	1	pep1	6	RRTSLGT	82.69
kin1	1	pep1	6	RAVSMDN	82.22
kin1	1	pep1	6	RRLTFRK	80.51
kin2	1	pep1	6	RHSSYPA	80.09