Assign columns to variables in shell

Question

I have this file:

NC_003037.1     453555  454448
NC_007493.2     2279220 2278345
NC_007952.1     1763831 1762950
NC_005791.1     844089  844916

I want to iterate of each line to obtain variables "id" "column1" "column2", that I will use in this command:

efetch -db nuccore -id $id -chr_start $column1 -chr_stop $column2 -format fasta > file.txt

Could you guide me with a way to do this in a shell script?

Answer 1

When you see multiple columns, think awk.

rm -f file.txt ; cat inputfile.txt | \
awk '{ print "efetch -db nuccore -id "$1" -chr_start "$2" -chr_stop "$3" -format fasta" }' | \
while read cmd; do $cmd >> file.txt; done

(EDIT: moved the re-direct into the execution)

The awk forms up the command, and then the while read cmd ...; done executes it.

Answer 2

Use a read loop:

while read -r id column1 column2; do
  efetch -db nuccore -id "$id" -chr_start "$column1" -chr_stop "$column2" -format fasta
done < file.input > file.txt

• it is important to use double quotes around variable expansions to prevent word splitting and globbing
• put > file.txt at the end of the loop for better I/O efficiency

---

Related:

• Looping through the content of a file in Bash
• When to wrap quotes around a shell variable?