Force script to finish even if it fails.

Question

[James]

Hi, Not sure I'm asking this the right way but does anyone know a way to get a bash script to finish even if it fails? I'm trying to automate running a script on all the illumina R1 and R2 files in a directory, Using 1 script to batch together the pair of reads.

For R1 in *R1*.gz
do 
Echo $R1
R2=`echo $R1 | sed 's/R1/R2/'`

MyPipeline.sh "$R1" "$R2"
Done

Mypipeline.sh then does denovo assembly, blasting a database to choose a reference, then mapping to that reference and creating a consensus sequence. My problem is that if the second script doesn't find a blast match it fails. The first script then doesn't move onto the next pair of raw reads. Please can anyone suggest a solution?

Thank you!

[Alex Reynolds]

What language are your scripts written in? There could be any number of ways to do this that are largely dependent on shell interpreter features. For example, a Javascript shell script could use Promises, or Python could use the finally clause in an exception handling block, and so on. Feel free to add a comment, or (preferably) edit your question to include more detail about what you're doing and how you're doing it.

[James]

Thanks for the reply. These are bash scripts. I've edited my post. I hope that's enough info?

[MSM55]

You can try following code.

total_files=`find -name '*.fastq' | wc -l`
arr=( $(ls *.fastq) )
for ((i=0; i<$total_files; i+=2))
{
sample_name=`echo ${arr[$i]} | awk -F "_" '{print $1}'`
echo " $sample_name"

MyPipeline.sh ${arr[$i]} ${arr[$i+1]}
 }

Answer 1

[Pierre Lindenbaum]

use a TRAP : http://redsymbol.net/articles/bash-exit-traps/

or use a OR operator to continue

ls file_not_exists || echo "CONTINUE"

Answer 2

[Alex Reynolds]

If you want to track that an error occurred, while still allowing the bash script to operate, then you could do this in the following way:

#!/bin/bash
set -e

for R1 in *R1*.gz
do 
  echo ${R1} >&2
  R2=`echo "${R1}" | sed 's/R1/R2/'`
  MyPipeline.sh "${R1}" "${R2}" >&2 || errorOccurred=true
done

if [ ${errorOccurred} ]
then
  echo "MyPipeline.sh failed somewhere -- check the error log!" >&2
  exit 1
fi

exit 0

If the script runs successfully, you'll get a 0 exit code — all went well:

$ ./someScript.sh 1> stdout.txt 2> stderr.txt
$ echo $?
0

If the script failed, you'll get a 1 exit code — failure:

$ ./someScript.sh 1> stdout.txt 2> stderr.txt
$ echo $?
1

The file stderr.txt is an "error log": a text file containing all the data piped to the standard error stream, useful for debugging.