Move just those files named as specific rows in a sample sheet - bash

Question

Imagine I have these files in my working directory in bash:
123.tsv 456.tsv 789.tsv 101112.tsv 131415.tsv
and that I have this sample sheet (tab separated):
sampleID tissue
123 lung
124 bone
456 lung
457 bone
Now, I want to move those files corresponding to lung samples to a new directory, so I would like to have the following files in the new directory:
123.tsv
456.tsv
I was trying to use:
awk -F"\t" '$2 == "lung"'
But I am not sure about how to include this in a for loop to select filenames included in the first column of the output file from the awk command.
How can I solve this?

Answer 1

If row number is larger 1 and second column contains lung then print content of first column with some text around it:
mkdir new_dir
awk 'NR>1 && $2=="lung" {print "mv", $1 ".tsv new_dir"}' sample.sheet
If output looks fine, append | sh to awk line to execute commands.

Answer 2

#!/bin/sh
#
#
me=$( basename "${0}" )
# Adjust these as needed. If you want to use your current
# working directory change (or remove) `/tmp/` to `./`.
old_dir="/tmp/foo"
new_dir="/tmp/bar"
list="/tmp/sample_sheet"
# Make sure all the pieces are available. Exit if not.
if [ ! -d "${old_dir}" ]
then
echo "ERROR: ${me}: Source '${old_dir}' does not exist." 1>&2
exit 1
elif [ ! -d "${new_dir}" ]
then
echo "ERROR: ${me}: Target '${new_dir}' does not exist." 1>&2
exit 2
elif [ ! -r "${list}" ]
then
echo "ERROR: ${me}: Sample sheet input '${list}' does not exist." 1>&2
exit 3
fi
# Iterate over the first column in `${list}`.
for file in $( awk 'NR>1 && $2=="lung" {print $1".tsv"}' "${list}" )
do
# If the file exists move it, if not do nothing.
if [ -f "${old_dir}/${file}" ]
then
echo "INFO: ${me}: mv ${old_dir}/${file} ${new_dir}/${file}"
mv "${old_dir}/${file}" "${new_dir}/${file}"
fi
done

Answer 3

Here's a script that you can run like, for example, this:
./move_files.sh lung
This works for both cases (lung and bone), and is general. Put this into a file called move_files.sh:
#!/usr/bin/env bash
files=$(sed -e "s/\([0-9]\{3\}\)\( *$1\)/\1/g" <(grep $1 eg.sheet))
if [ ! -d $1 ]; then
mkdir $1
fi
for t in ${files[#]}; do
mv "./$t.tsv" $1
done
With the following directory content:
101112.tsv 123.tsv 124.tsv 131415.tsv 456.tsv 457.tsv 789.tsv eg.sheet move_files.sh
and eg.sheet containing:
sampleID tissue
123 lung
124 bone
456 lung
457 bone
... running the script with
./move_files.sh lung
... results in 123.tsv and 456.tsv being moved into a newly created lung directory (or simply moved there if the directory already exists).
You can then simply run
./move_files.sh bone
to move 124.tsv and 457.tsv to a newly created bone directory. Of course this is then generalisable to whatever is in eg.sheet.
Side note: you must run chomd +x move_files.sh in order to use it in the way I've suggested. Otherwise, you can invoke it with bash move_files.sh lung instead.
EDIT:
To address the point raised by keithpjolley in the comments, this can still work with "tissues" such as "eye lash" just by quoting the $1 variable throughout and by calling it with a quoted string (e.g., ./move_files.sh "eye lash"):
#!/usr/bin/env bash
files=$(sed -e "s/\([0-9]\{3\}\)\( *$1\)/\1/g" <(grep "$1" eg.sheet))
if [ ! -d "$1" ]; then
mkdir "$1"
fi
for t in ${files[#]}; do
mv "./$t.tsv" "$1"
done