looping sed commands

[tonybert]

greatings, I was wondering if anyone could provide any clues/assitance in automating a series of commands. Currently, I am formatting Illumina data for a UPARSE/USEARCH pipeline. Post quality filtered, trimmed samples need to have the fasta header changed to from
">Sample1_sequence1"

to

">barcodelabel=Sample1;Sample1_sequence1"

This is easy to do with sed;

sed 's/>Sample1/barcodelabel=Sample1/g' sample.fna > sample.rename.fna

However, the file I have has 48 different samples, and I was wondering if there is way to automate or loop sed, so that it might be able to read the Sample names from a text file, and then replace them in the new file.

Any suggestions are welcome. Thanks,

-tony

[WhatsOEver]

Use the following commands to loop either linewise through or through all files of a certain directory. From that you can use e.g. "basename" to get the filename without the path or parse it in any other way you need it.

In order to loop linewise through a file, use:

#Set the field seperator (IFS) to a line break
IFS="
"
for line in `cat yourSampleNameFile.txt`;do
sampleName = $line
...
done

If you want to loop through all files in a certain directory, use:

FILES=/path/to/your/sample/files/*
for f in $FILES
do
echo "Processing file $f..."
...
done

[tomc]

sed 's/^>\([^_]*\)_/>barcodelabel=\1;\1_/g' sample.fna > sample.rename.fna

Should work for any number of sample names as long as they are
between a leading ">" and followed with underscore.
one pass, no redundant external copy of sample names required.

Pitfalls to watch out for are lines with the start & end pattern
but are not sample definition lines and empty sample names.
These issues can be addressed by creating a more specific pattern recognizer
within "\([^_]*\)" which is currently just "accept anything but an underscore".