alexpreynolds
commented
3 years ago - I might also suggest using double-quotation marks around variables that are at risk of containing space characters. Using double-quotes will preserve spaces, but without them, a filename could become two or more filenames.
Lines 20-22 could be vulnerable to that, if the BIOSAMPLE name from the first column of METADATA might contain sample names with any spaces. Though less risky, GENOME and CHR could also be vulnerable, if they are provided with spaces, for whatever reason (user error, etc.).
jacobquon
meuleman
The following script may not be obsolete:
https://github.com/meuleman/epilogos/blob/80bf1f5b0846b495d40eda0cd0315c8c0d371b2c/scripts/preprocess_data_ChromHMM.sh#L1-L33
Some general suggestions that I have are:
Use
set -ueo pipefailbelow the shebang line to ensure variables are set and to handle error states more closely. You could even add an-xto log each command. This can be useful for pipelines, log statements, or debugging a problematic script.Try to avoid using uppercase variable names in bash scripts. Some variables in bash are reserved POSIX names and your intended values will get clobbered, if you accidently use these variable names. It can lead to non-obvious and frustrating errors.
Use curly braces around positional variables like
$1to${1},$2to${2}etc. Not a big deal, here, but a good habit to get into as this allows you to specify more than nine input parameters ($10will not work, but${10}will, for example).