Part 5 of "Introduction to Linux for Bioinformatics": Working the command line's text tools

This presentation is available under the Creative Commons
Attribution-ShareAlike 3.0 Unported License. Please refer to
http://www.bits.vib.be/ if you use this presentation or parts
hereof.
Introduction to Linux
for Bioinformatics
Working the command line
Joachim Jacob
5 and 12 May 2014

2 of 81
Short recapitulation of last week
Bash only looks at certain directories for
commands/software/programs …
The location of a tool you can find with 'which'.

3 of 81
We can install and run software
● E.g. commands for mapping NGS data on the wiki:
http://wiki.bits.vib.be/index.php/GenomeView_Workshop:_Mapping_excerc
ises#Mapping

4 of 81
Software installation directories
Contain the commands we can execute in the terminal

5 of 81
Software installation directories
How does the terminal know where to look for
executables? (e.g. how does it know bowtie is
in /usr/bin?)

6 of 81
Environment variables
A variable is a word that represents/contains a
value or string. Environment variables describe
your system. Fictive example:
TREES are green on
my system
Program that
Draws trees.
TREES=green
Environment variable

7 of 81
Programs use env variables
TREES=purple
TREES = purple
on my system
Program that
Draws trees.
Depending on how environment variables are set,
programs can change their behaviour.

8 of 81
'env' displays environment vars

9 of 81
Programs need to be in the PATH
● Environment variables: they contain
values, applicable system wide.
●
●
https://help.ubuntu.com/community/EnvironmentVariables

10 of 81
The PATH environment variable
PATH contains a set of directories, separated by ':'
$ echo $PATH
/home/joachim/bin:/usr/local/sbin:/usr/local/bin:/usr/
sbin:/usr/bin:/sbin:/bin:/usr/games

11 of 81
Installing is just placing the executable
1. You copy the executable to one of the folders
in PATH
2. You create a sym(bolic) link to an executable
in the one of the folders in PATH
(see previous week)
3. You add a directory to the PATH variable
$ sudo cp /home/joachim/Downloads/tSNE /usr/local/bin
$ sudo chmod +x /usr/local/bin/tSNE

12 of 81
3. Add a directory to the PATH
Export <environment_variable_name>=<value>

13 of 81
Env variables are stored in a text file
$ cat /etc/environment
PATH="/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin
:/bin:/usr/games"
/etc is the directory that contains configuration text
files. It is only owned by root: system-wide settings.
A 'normal' user (session-wide settings) can create the
file ~/.pam_environment to set the vars with content

14 of 81
Recap: editing files
Create a text file with the name .pam_environment
and open in an editor:
$ nano .pam_environment
→ quit by pressing ctrl-x
$ gedit .pam_environment
→ graphical
$ vim .pam_environment
→ quit by pressing one after the other :q!

15 of 81
Create .pam_environment
In ~/.pam_environment, type:
TREES DEFAULT=green
Save the file. Log out and log back in.

16 of 81
Bash variables are limited in scope
You can assign any variable you like in bash, like:
The name of the variable can be any normal string.
This variable exists only in this terminal. The
command echo can display the value assigned to that
variable. The value of a variable is referred to by $
{varname} or $varname.

17 of 81
It can be used in scripts!
All commands you type, you can put one after the
other in a text file, and let bash execute it.
Let's try!
Make a file in your ~ called 'space_left':
Enter two following bash commands in this file:
df -h .
du -sh */

18 of 81
Running our first bash script

19 of 81
The shebang
Simple text files become Bash scripts when adding
a shebang line as first line, saying which program
should read and execute this text file.
#!/bin/bash
#!/usr/bin/perl
#!/usr/bin/python
(see our other trainings for perl and python)

20 of 81
Things to remember
● Linux determines files types based on its
content (not extension).
● Change permissions of scripts to read and
execute to allow running in the command line:
$ chmod +x filename

21 of 81
Exercise
→ A simple bash script

22 of 81
Can you reconstruct the script?
.......................................
.......................................
.......................................
.......................................

23 of 81
One slide on permissions
$ chown user:group filename
$ chmod [ugo][+-][rwx] filename
or
$ chmod [0-7][0-7][0-7] filename
1 stands for execute
2 stands for write
4 stands for read
→ any number from 0 to 7 is a unique
combination of 1, 2 and 4.

24 of 81
Passing arguments to bash scripts
We can pass on arguments to our scripts: they are
subsequently stored in variables called $1, $2, $3,...
Make a file called 'arguments.sh' with following
contents (copy paste is fine – be aware of the “):
#!/bin/bash
firstarg=$1
secondarg=$2
echo “You have entered ”$firstarg”
and ”$secondarg””

25 of 81
Make your script executable.
$ chmod +x arguments.sh

26 of 81
Let's try to look at it, and run it.

27 of 81
Arguments are separated by white spaces
The string after the command is chopped on the
white spaces. Different cases (note the “ and ):

28 of 81
Arguments are separated by white spaces
The string after the command is chopped on the
white spaces. Different cases (note the “ and ):

29 of 81
Useful example in bioinformatics
For example, look at the script on our wiki:
http://wiki.bits.vib.be/index.php/Bfast
Lines starting with # are ignored.

30 of 81
Chaining command line tools
This is the ultimate power of Unix-like OSes. The
philosophy is that every tool should do one
small specific task. By combining tools we can
create a bigger piece of software fulfilling our
needs.
How combining different tools?
1. writing scripts
2. pipes

31 of 81
Chaining the output to input
What the programs take in, and what they print out...

32 of 81
We can take the output of one program, store it,
and use it as input for another program
~ Assembly line

33 of 81
Deliverance through channels
When a program is executed, 3 channels are opened:
● stdin: an input channel – what is read by the program
● stdout: channel used for functional output
● stderr: channel used for error reporting
In UNIX, open files have an identification number
called a file descriptor:
● stdin called by 0
● stdout called by 1
● stderr called by 2
(*) by convention
http://www.linux-tutorial.info/modules.php?name=MContent&pageid=2
1

34 of 81
I/O redirection of terminal programs
$ cat --help
Usage: cat [OPTION]... [FILE]...
Concatenate FILE(s), or standard input,
to standard output.
“When cat is
run it waits
for input. As
soon as an
enter is entered
output is written
to STDOUT.”
cat
STDIN
or channel 0
STDOUT
or channel 1
STDERR
or channel 2

35 of 81
I/O redirection
When cat is launched without any arguments, the
program reads from stdin (keyboard) and writes to
stdout (terminal).
Example:
$ cat
type:
DNA: National Dyslexia Association↵
result:
DNA: National Dyslexia Association
You can stop the program using the 'End Of Input'
character CTRL-D

36 of 81
Takes input from the keyboard, attached to
STDIN.
cat
1 2
0
STDIN
or channel 0
STDOUT
or channel 1
STDERR
or channel 2

37 of 81
Takes input from files, which is attached to
STDIN
cat
1 2
0
STDERR
or channel 2
STDOUT
or channel 1

38 of 81
Connect a file to STDIN:
$ cat 0< file
or shorter:
$ cat < file
or even shorter (and most used – what we know
already)
$ cat file
Example:
$ cat ~/arguments.sh
Try also:
$ cat 0<arguments.sh

39 of 81
Output redirection
Can write output to files, instead of the
terminal
cat
1 2
0
STDERR
or channel 2
STDOUT
or channel 1

40 of 81
Output redirection
 The stdout output of a program can be
saved to a file (or device):
$ cat 1> file
or short:
$ cat > file
 Examples:
$ ls -lR / > /tmp/ls-lR
$ less /tmp/ls-lR

41 of 81
You have noticed that running:
$ ls -lR / > /tmp/ls-lR
outputs some warnings/errors on the screen: this
is all output of STDERR (note: channel 1 is
redirected to a file, leaving only channel 2 to the
terminal)

42 of 81
Redirect the errors to a file called 'error.txt'.
$ ls -lR /
?

43 of 81
Redirect the error channel to a file error.txt.
$ ls -lR / 2 > error.txt
$ less error.txt

44 of 81
Beware of overwriting output
IMPORTANT, if you write to a file, the contents
are being replaced by the output.
To append to file, you use:
$ cat 1>> file
or short
$ cat >> file
Example:
$ echo “Hello” >> append.txt
$ echo “World” >> append.txt
$ cat append.txt

45 of 81
INPUT OUTPUT
Input from file < filename > filename Output to a file
Input until
string EOF
<< EOF >> filename Append output to a
file
Input directly
from string
<<< “This string is read”

46 of 81
Special devices
● For input:
/dev/zero all zeros
/dev/urandom (pseudo) random numbers
● For output:
/dev/null 'bit-heaven'
Example:
You are not interested in the errors from the a
certain command: send them to /dev/null !
$ ls -alh / 2> /dev/null

47 of 81
Summary of output redirection
● Error direction to /dev/null
● The program can run on its
own: input from file, output to
file.
cat
1 2
0
/dev/null

48 of 81
Plumbing with in- and outputs
● Example:
$ ls -lR ~ > /tmp/ls-lR
$ less < /tmp/ls-lR ('Q' to quit)
$ rm -rf /tmp/ls-lR

49 of 81
Plumbing with in- and outputs
● Example:
$ ls -lR ~ > /tmp/ls-lR
$ less < /tmp/ls-lR ('Q' to quit)
$ rm -rf /tmp/ls-lR
can be shortened to:
$ ls -lR ~ | less
(Formally speaking: the stdout channel of ls is connected to the
stdin channel of less)

50 of 81
Combining pipe after pipe
Pipes can pass the output
of a command to another
command, which on his
turn can pipe it through,
until the final output is reached.
$ history | awk '{ print $2 }'
| sort | uniq -c | sort -nr | head -3
237 ls
180 cd
103 ll

51 of 81
Text tools combine well with pipes
UNIX has an extensive toolkit for text analysis:
● Extraction: head, tail, grep, awk, uniq
● Reporting: wc
● Manipulation: dos2unix, sort, tr, sed
But, the UNIX tool for heavy text parsing is perl (see
https://www.bits.vib.be/index.php/training/175-perl)

52 of 81
Grep: filter lines from text
grep extracts lines that match a string.
Syntax:
$ grep [options] regex [file(s)]
The file(s) are read line by line. If the line matches
the given criteria, the entire line is written to
stdout.

53 of 81
Grep example
A GFF file contains genome annotation
information. Different types of annotations are
mixed: gene, mRNA, exons, …
Filtering out one type of annotation is very easy
with grep.
Task:
Filter out all lines from locus Os01g01070 in
all.gff3 (should be somewhere in your Rice folder).

55 of 81
Grep
-i: ignore case
matches the regex case insensitively
-v: inverse
shows all lines that do not match the regex
-l: list
shows only the name of the files that contain a
match
-n:
shows n lines around the match
--color:
highlights the match

56 of 81
Finetuning filtering with regexes
A regular expression, aka regex, is a formal way of
describing sets of strings, used by many tools: grep,
sed, awk, ...
It resembles wild card-functionility (e.g. ls *) (also
called globbing), but is more extensive.
http://www.faqs.org/docs/abs/HTML/regexp.html
http://www.slideshare.net/wvcrieki/bioinformatics-p2p3perlregexes-v2013wimvancriekinge?utm_source=slideshow&utm_medium=ssemail&u

57 of 81
Basics of regexes
A plain character in a regex matches itself.
. = any character
^ = beginning of the line
$ = end of the line
[ ] = a set of characters
Example:
$ grep chr[1-5] all.gff3
Regex chr chr[1-5] chr. AAF12.[1-3] AT[1,5]G[:digit:]+/.[1,2]
Matching
string set
chr1 chr1 chr1 AAF12.1 AT5G08160.1
chr4 chr4 chr4 AT1G14525.1
chr5 chr5 chr5

58 of 81
Basics of regexes
Example: from TAIR9_mRNA.bed, filter out the mRNA
structures from chr1 and only on the + strand.
$ egrep '^chr1.++' TAIR9_mRNA.bed > out.txt
^ matches
the start of
a string
^chr1
Matches lines
With 'chr1' appearing
At the beginning
. matches
any char
.+
matches
any string
Since + is a special character
(standing for a repeat of one or more),
we need to escape it.
+ matches a '+' symbol as such
Together in this order, the regex
filters out lines of chr1 on + strand

59 of 81
Finetuning filtering with regexes
$ egrep '^chr1.++' TAIR9_mRNA.bed > out.txt
chr1 2025600 2027271 AT1G06620.10 + 2025617 2027094 0 3541,322,4
chr1 16269074 16270513 AT1G43171.10 + 1626998816270327 0
chr1 28251959 28253619 AT1G75280.10 + 2825202928253355 0
chr1 693479 696382 AT1G03010.10 + 693479 696188 0 592,67,119
http://www.faqs.org/docs/abs/HTML/regexp.html
http://www.slideshare.net/wvcrieki/bioinformatics-p2p3perlregexes-v2013wimvancriekinge?utm_source=slideshow&utm_medium=ssemail&u

60 of 81
Wc – count words in files
A general tool for counting lines, words and characters:
wc [options] file(s)
c: show number of characters
w: show number of words
l: show number of lines
How many mRNA entries are on chr1 of A. thaliana?
$ wc -l chr1_TAIR9_mRNA.bed
or
$ grep chr1 TAIR9_mRNA.bed | wc -l

61 of 81
Translate
To replace characters:
$ tr 's1' 's2'
! tr always reads from stdin – you cannot specify any
files as command line arguments. Characters in s1
are replaced by characters in s2.
Example:
$ echo 'James Watson' | tr '[a-z]' '[A-Z]'
JAMES WATSON

62 of 81
Delete characters
To remove a particular set of characters:
$ tr -d 's1'
Deletes all characters in s1
Example:
$ tr –d 'r' < DOStext > UNIXtext

63 of 81
awk can extract and rearrange
… specific fields and do calculations and
manipulations on it.
awk -F delim '{ print $x }'
● -F delim: the field separator (default is white space)
● $x the field number:
$0: the complete line
$1: first field
$2: second field
…
NF is the cumber of fields (can also be taken for last
field).

64 of 81
For example: TAIR9_mRNA.bed needs to be converted to
.gff (general feature format). See the .gff format
http://wiki.bits.vib.be/index.php/.gff
With AWK this can easily be done! One line of .bed looks
like:
→ needs to be one line of .gff
chr1 2025600 2027271 AT1G06620.10 + 2025617 2027094 0 3541,322,429, 0,

65 of 81
$ awk '{print $1”tawktmRNAt”$2”t”$3”t”
$5”t”$6”t0t”$4 }' TAIR9_mRNA.bed
chr1 2025600 2027271 AT1G06620.10 + 2025617 2027094 0 3541,322,429, 0

66 of 81
Awk has also a filtering option
Extraction of one or more fields from a tabular data
stream of lines that match a given regex:
awk -F delim '/regex/ { print $x }'
Here is:
● Delim: the delimiter in the file
● regex: a regular expression
The awk script is executed only if the line matches
regex lines that do not match regex are removed
from the stream
Further excellent documentation on awk:
http://www.grymoire.com/Unix/Awk.html

67 of 81
Cut selects columns
Cut extracts fields from text files:
● Using fixed delimiter
$ cut [-d delim] -f <fields> [file]
● chopping on fixed width
$ cut -c <fields> [file]
For <fields>:
N the Nth element
N-M element the Nth till the Mth element
N- from the Nth element on
-M till the Mth element
The first element is 1.

68 of 81
Cutting columns from text files
Fixed width example:
Suppose there is a file fixed.txt with content
12345ABCDE67890FGHIJ
To extract a range of characters:
$ cut -c 6-10 fixed.txt
ABCDE

69 of 81
Sorting output
To sort alphabetically or numerically lines of text:
$ sort [options] file(s)
When more files are specified, they are read one by
one, but all lines together are sorted.

70 of 81
Sorting options
● n sort numerically
● f fold – case-insensitive
● r reverse sort order
● ts use s as field separator (instead of space)
● kn sort on the n-th field (1 being the first field)
Example: sort mRNA by chromosome number and
next by number of exons.
$ sort -n -k1 -k10 TAIR9_mRNA.bed >
out.bed

71 of 81
Detecting unique records with uniq
● eliminate duplicate lines in a set of files
● display unique lines
● display and count duplicate lines
Very important: uniq always needs from sorted
input.
Useful option:
-c count the number of fields.

72 of 81
Eliminate duplicates
● Example:
$ who
root tty1 Oct 16 23:20
james tty2 Oct 16 23:20
james pts/0 Oct 16 23:21
$ who | awk '{print $1}' | sort | uniq
james
root

73 of 81
Display unique or duplicate lines
● To display lines that occur only once:
$ uniq -u file(s)
● To display lines that occur more than once:
$ uniq -d file(s)
Example:
$ who|awk '{print $1}'|sort|uniq -d
james
● To display the counts of the lines
$ uniq -c file(s)
Example
$ who | awk '{print $1}' | sort | uniq -c
4 james
1 root

74 of 81
Edit per line with sed
Sed (the stream editor) can make changes in text
per line. It works on files or on STDIN.
See http://www.grymoire.com/Unix/Sed.html
This is also a very big tool, but we will only look to
the substitute function (the most used one).
$ sed -e 's/r1/s1/' file(s)
s: the substitute command
/: separator
r1: regex to be replaced
s1: text that will replace the regex match

75 of 81
Paste several lines together
Paste allows you to concatenate every n lines into one
line, ideal for manipulating fastq files.
We can use sed for this together with paste.
$ paste - - - - < in.fq |
cut -f 1,2 |
sed 's/^@/>/' |
tr "t" "n" > out.fa
http://thegenomefactory.blogspot.be/2012/05/cool-use-of-unix-paste-with-ngs.html

76 of 81
Transpose is not a standard tool
http://sourceforge.net/projects/transpose/
But it is extremely useful. It transposes tabular text
files, exchanging columns for row and vice versa.

77 of 81
Building your pipelines
grep
tr
sed awk
sort
wc uniq
uniq
transpose
paste

78 of 81
Fill in the tools
Filter on lines ánd select different columns: …....................
Merge identical
fields: …...........
Filter lines: …............
Sort lines: ….....
Select columns: …............
Replace characters: ….....
Replace strings: …............
Transpose: ….....Numerical summary: ….....

79 of 81
Exercise
→ Text manipulation exercises

80 of 81
Keywords
Environment variable
PATH
shebang
script
argument
STDIN
pipe
comment
Write in your own words what the terms mean

Part 5 of "Introduction to Linux for Bioinformatics": Working the command line's text tools

More Related Content

What's hot (20)

Viewers also liked (17)

Similar to Part 5 of "Introduction to Linux for Bioinformatics": Working the command line's text tools (20)

More from Joachim Jacob (8)

Recently uploaded (20)

Part 5 of "Introduction to Linux for Bioinformatics": Working the command line's text tools