Shell Programming and Scripting

I have a file from which I want to eliminate the numerical values..
The contents of the file are as shown below..

Code:

1    a1,b,2    1,b,c
2    a2,4,b    a,b,2

From the above file I want eliminate only the numberical values(except the line numbers which are at the beginning)..

The file consists of a line number, followed by a tab, followed by two columns with a tab seperating them!

I've used the sed command without success..
Please take a look at my code..
I'm trying to replace a number with a comma (,)
Please note: I don't want to eliminate the digit '1' in the variable "a1".. I need to eliminate only pure numbers..

Code:

sed 's/,[0-9]+[,]*/,/g
s/,[0-9]+[\t]/,/g' file.txt

But its not working!! Please help me!

Show us desired output for that sample data.

Ya here it goes..

Code:

1    a1,b,,    ,,b,c
2    a2,,,b    a,b,,

The numbers should be replaced with a semi colon(,)..

Code:

sed ':b s/\(,[^0-9]*\)[0-9]/\1,/g;t b' infile

or

Code:

sed ':b s/\(,[^[:digit:]]*\)[[:digit:]]/\1,/g;t b' infile

Tested successfully on linux / GNU sed.

You may get a "label too long" error on Sun sed.
i didn't test on other plateform.

A perl:

Code:

perl -pne 's/[^^](\b)\d[,\s]/\1,,/g' file

Hey its not working for the file given below..

Thank u so much!!
Its working!

But I'm a noob.. could you please tell me what is ":b" "t" and "b" doing in the script??