What Are The Bioinformatics-Related Aliases Or Functions In Your Bashrc

Question

[brentp]

I've added a couple of things to my bashrc that I use pretty often. Some of the simpler ones:

greatly reduce time for a lot of operations.

 export LC_ALL=C

e.g sorting a 1.8m line bed file goes from 43 seconds with LC_ALL="" to 3.2 seconds with LC_ALL=C

A quick check to make sure all lines have the same number of columns:

function check-columns(){
    awk 'BEGIN {FS="\t" }{ print NF }' $1 | sort -u
}

Output tab-delimited output so that the columns are aligned:

alias cols="column -s$'\t' -t"

use like:

head some.bed | cols

[Stefano Berri]

Nice question. Looking forward for answers. I keep on bumping in this LC_ALL thing. Do you have a nice link that explain what locale is andwhy it matters?

[brentp]

IIUC LC_ALL=C tells whoever will listen that strings are not multi-byte (e.g. unicode), so no conversion is needed.

Answer 1

[Chris Miller]

A few things from mine:

# easy way to do things per-chromosome:
# for i in $CHROM;do *your code*;done
export CHROM="$(seq 1 22) X Y MT"

alias sv='samtools view'
alias svh='samtools view -h'

#grab the header of a VCF
function vcfhead {
    head -n 1000 $1 | grep "^#"
}

#unwad tarballs
alias unwad='tar -xzvf'

#find in the current directory
alias ff='find . -name $1'

#sum a column of integers
alias sumcol='awk '\''{ SUM += $1} END { print SUM}'\'

#sum a column of floats
alias sumcolfloat='awk '\''{ SUM += $1} END { OFMT="%4.2f"; print SUM}'\'' <$i'

#convert csv to tab-delimited
alias csv2tab='sed '\''s/\,/\t/g'\'

# convert tabs to new lines
alias tab2nl='perl -pe "s/\t/\n/g"'

#git stuff
#shows a preview of what's outgoing if you do a git push
function grout {
 git fetch origin master
 gd2 $(parse_git_branch) FETCH_HEAD
}
#shows a preview of what's incoming if you do a git pull
function grin {
 git fetch origin master
 gd2 FETCH_HEAD $(parse_git_branch)
}

#show the column headers with corresponding field numbers
#I use this constantly
alias header='head -n 1 | tab2nl | cat -n'

[Frédéric Mahé]

Just for the fun, here is a Awk replacement for your "header" command:

awk 'BEGIN {FS="\t"; OFS="\t"} NR==1 {for (i=1; i <= NF; i++) {print i, $i}}'

[Ryan D]

This was useful since the header alias didn't work in my .bashrc alias set (even with the tab2nl alias added). I made this a function. Agree with Madeline, though. Really useful.

[Madelaine Gogol]

Oh, I like your header / tab2nl combo.

Answer 2

[Jeremy Leipzig]

Not to beat this to death but I find the directory-based bash history scheme indispensable. I honestly don't know how people function with global histories.

function mycd()
{
    tmpDir="$PWD"
    echo "#"`date +%s`" $USER -> $@"  >> "$HISTFILE"

    builtin cd "$@" # do actual cd                                                                        

    #if this directory is writable then write to directory-based history file
    #otherwise write history in the usual home-based history file                    
    touch "$PWD/.dir_bash_history" 2>/dev/null && export HISTFILE="$PWD/.dir_bash_history" || export HISTFILE="$HOME/.bash_history";
    echo "#"`date +%s`" $USER <- $OLDPWD" >> "$HISTFILE"
}
alias cd="mycd"
#initial shell opened                                                                                     
export HISTFILE="$PWD/.dir_bash_history"
#timestamp all history entries                                                                            
export HISTTIMEFORMAT="%h/%d - %H:%M:%S "
export HISTCONTROL=ignoredups:erasedups
export HISTSIZE=1000000
export HISTFILESIZE=1000000
shopt -s histappend ## append, no clearouts                                                               
shopt -s histverify ## edit a recalled history line before executing                                      
shopt -s histreedit ## reedit a history substitution line if it failed                                    

## Save the history after each command finishes                                                           
## (and keep any existing PROMPT_COMMAND settings)                                                        
export PROMPT_COMMAND="history -a; history -c; history -r; $PROMPT_COMMAND"

[Madelaine Gogol]

Okay, okay, I'll do it. Thanks.

[Istvan Albert]

my thoughts exactly ;-)

[brentp]

accepting this answer because of the directory-specific history. really useful.

Answer 3

[Madelaine Gogol]

Thanks for the tips. My suggestion might be obvious, but I generally want to find something in the same set of directories, so it's a speedup for me:

Seach R scripts in a bunch of directories

alias searchR='find /place1 /place2 -name "*.R" | xargs grep '

A few of the answers from here apply: http://www.biostars.org/post/show/6660/bioinformatics-cheat-sheet/

I have a lot of little one-liners and snippets that I don't use often enough that I've bothered to alias them, though. These I have tended to save in text files in a common directory in files called awktips.txt, sedtips.txt, and I just go and grab the line of interest as needed. Perhaps I should go through the trouble to alias some of them, but I'd probably forget the alias name I gave it anyway and have to look it up.

[jake.biesinger]

Wouldn't ack --R pattern do the same as your searchR alias?

[Madelaine Gogol]

Oh... I hadn't heard of ack. Looks like it does cwd on down, though, not multiple directories?

Answer 4

[Istvan Albert]

Heng Li's bioawk linked as hawk ;-) and virtualenv activation.

I do add lots of short term aliases when working with a specific tool as well.

Answer 5

[Andreas]

Just to add two more:

Get number of sequences in a (gzipped) fasta file:

alias fasta_num_seq='zgrep -c "^>"'

Get the number of reads from a (gzipped) fastq file (note: in theory '+' could also happen as first character in the quality line):

alias fastq_num_seq='zgrep -c "^+"'