UNIX.md

UNIX Command-line Guide

These are some of the more commonly used UNIX commands.

BEGINNER commands

scroll through your command history

<up/down arrow>

kill the current process (or delete what you have written on the command line)

<control>-c

change into your “home” directory

cd

Find out what directory you are in.

pwd

list the contents of the current directory/folder

ls

list the contents of some other directory

ls /path/to/other/directory

make a new directory (aka folder) for your programs (binaries)

mkdir bin

make sure it got created

ls

list the contents of a directory

ls bin

enter into the “bin” directory (cd means "change directory")

cd bin

exit the "bin" directory

(change back to the "parent" directory, the one just above this in the file system)

cd ..

remove a file

rm unwantedfile.txt

remove a directory

rm -r unwanteddirectory

INTERMEDIATE commands

unzip a file

unzip zippedfile.zip

list the number of lines, words, and characters in a file:

wc filename.txt

view the first 10 lines of a file

head myfile.txt

print an entire file to the screen (might be long)

cat filename.txt

view the first 3 columns (fields) of a file, assuming tab delimiters

cut -f 1-3 myfile.txt

view the first 3 columns (fields) of a file, assuming comma delimiters

cut -f 1-3 -d "," myfile.txt

view the first 10 lines of the first 3 columns

(note the use of | to pipe the output of one command to another)

cut -f 1-3 myfile.txt | head 

view the first 20 characters of each line in a file

cut -c 1-20 myfile.txt

view the first 20 characters of the first 10 lines in a file

cut -c 1-20 myfile.txt | head 

scroll through a file (space bar jumps down a page, "b" jumps back a page)

less myfile.txt

scroll through the first 3 columns of the whole file (enter q to exit)

(note again the use of | to pipe the output of cut to less)

cut -f 1-10 -d "," myfile.txt | less

make a copy of a file

cp file.txt filecopy.txt

copy a file to a directory

cp file.txt directory_name

copy a distant file to the current directory (. means current directory)

cp /full/path/to/other/file.txt .

make a copy of a directory

cp -r directory1 directory2

mv (rename) a file

mv oldname.txt newname.txt

see how large files are

du -hs *

edit a file (exit with ctrl-x)

nano somefile.txt

search a file for a string or word

(e.g. print out all lines that have a ">" character)

grep ">" input.fna

search for lines that do not match a pattern

(e.g. print lines that DO NOT have a ">" character)

grep -v ">" input.fna

count all lines that have a ">" character:

grep -c ">" input.fna

ADVANCED commands

Using "Screen" to keep a session open

If you need to run something that will take a while, there are ways to keep an interactive computing session open on MSI. One way is to log in to a specific login node (e.g. ahl01), open a "screen" session, and then start your interactive session. The screen session will stay open on the login node if you disconnect and connect again later. Here are the steps:

1. When you first connect to MSI, connect to a specific login node:

ssh username@ahl01.msi.umn.edu

2. Start a "screen" session (like opening a browser window, conceptually)

screen

3. Launch your compute node

srun ...

4. When you need to log off temporarily, detach from the screen session (don't hold ctrl after the a)

<ctrl>-a d

Then exit the login node with <ctrl>-d or exit. 5. When ready to log back on, connect to the same login node:

ssh username@ahl01.msi.umn.edu

6. Re-attach to the screen session

screen -Dr

Now you are back on the compute node, right where you left off. If you had a command running, it will have continued running in the background.

Other helpful screen commands

start another screen "window" (like opening another tab in a browser)

<ctrl>-a c

move to the next screen "tab"

<ctrl>-a n

move to the previous screen "tab"

<ctrl>-a p

terminate/permanently close a commandline session or screen session

<ctrl>-d

search your command history

<control>-r (then type the search string)

clear your screen

clear