Shell Programming and Scripting

Hello!

Can anybody suggest about the fastest way of extracting "n" random columns from a very large file (tab separated) having thousands of columns, where n can be any specified number.

Thanks!

Just one line of thousands of columns?

If you have or can use the rl utility, it is quite fast indeed:

Code:

rl -d $'\t' -c 32 tabseperatedfile

You can use the "cut" command - the default delimiter is horizontal tab:

Code:

cut -fn Inp_File

Where "n" is the column number which can be of any length, including zero.

Hey! may be I was not very clear. I wanted to extract 'n' random columns where 'n' can be 250 or 400 or any other number. And the file has thousands of columns (lets say 5000) and hundreds of rows (lets say 400).

Thanks!

Actually it is not very clear

lets say you give n=100, you want to extract first 100 columns from all the rows.

regards,
Ahamed

---------- Post updated at 02:12 AM ---------- Previous update was at 02:06 AM ----------

To extract columns from m to n where n>m

Code:

#awk '{gsub("\t"," ",$0);print}' file | cut -d" " -fm-n
#eg:
awk '{gsub("\t"," ",$0);print}' file | cut -d" " -f3-5

#eg to extract first 400 columns
awk '{gsub("\t"," ",$0);print}' file | cut -d" " -f1-400

regards,
Ahamed

Thanks for your reply ahamed but I need to extract random columns like if I say I want to extract 50 random columns it should extract it randomly from all the columns, lets say it extracts the column no. 10, 28, 40, 57, 92, 500, 740, 2540, ... or any other columns for that matter but it should be random and not ordered.

Ofcouse the total no. of columns and rows are fixed in that file and 'n' the number of random columns required can vary.

Ruby(1.9.1+)

Code:

f=open("file")
numcols=f.readline.split.size
cols=(1..numcols).to_a.sample(4)
while not f.eof?
    line=f.gets.split
    cols.each{|x| printf "%s " , line[x-1]}
    puts
end
f.close