How to replace a part of a string using bash?

string="1,2,3"
names=( nobody tom jerry sue )
indexes=( $(sed 's/,/ /g' <<< "$string") )
result=""
pad=""
for i in ${indexes[@]}
do
    result="$result$pad${names[$i]}"
    pad=","
done
echo "$result"

Explanation:

1. Simulate input string from user
2. Array of names, indexed from 0 (nobody) to 3 (sue)
3. Array of numbers generated from string by requesting sed to replace commas by blanks. There are other ways to do that, but I tend to use sed still. For example, indexes=( ${string//,/ } will do the job without invoking sed or any external program.
4. The output goes in result.
5. I use pad to build up strings with a separator.
6. Loop over the numbers in the array indexes
7. Result is the concatenation of what was there before, the padding, and the name represented by the number in $i.
8. Set the pad to a comma for the second and subsequent iterations.
9. Echo the result.

I didn't count the 'do' or 'done' lines.

I should've been more specific and asked you how the sed syntax works.

There are 4 interesting parts to the line:

indexes=( $(sed 's/,/ /g' <<< "$string") )

1. indexes=( ... )

   This is an array assignment. It will take the words produce by whatever is in ... and treat them as the elements of the array. That's how the names=( ... ) assignment works, of course.

2. sed ... <<< "$string"

   This runs the sed command with the value of "$string" as its standard input. This is a bash extension called a Here String. Thus the only line of input is "1,2,3".

3. 's/,/ /g' is a sed command that substitutes a space for every comma in the input. Thus, the output is a line containing 1 2 3.

4. $( ... ) runs the command contained within the parentheses (the sed command) and treats the output as a string.

The overall effect is the same as:

indexes=( 1 2 3 )

The spaces around the parentheses aren't strictly necessary, but I think they make the array assignment easier to spot — but it is a notation I'm still 'learning' (having to remember to use).