Syntax error with FOR nested inside IF in bash.

Question

Im trying to make this pipeline more flexible. So say I want to be able to easily switch the number of times a for loop is run, based on whether or not I want to analyse all my data.

#!/bin/bash
#AllCohorts='Yes'
AllCohorts='NotAll'
groups=$(wc -l Groups | awk '{print $1}' ) 

if [[ $AllCohorts == *Yes* ]]
then
    for pheno in `seq 1 $groups`; 
    do
    out=$(sed -n "$pheno"'p' Groups) 
    whichPheno=$pheno
elif [[ $AllCohorts == *Not* ]]
then
    nbGroups=$(wc -l TestingGroups | awk '{print$1}')
    for pheno in `seq 1 $nbGroups`; 
    do
    out=$(sed -n "$pheno"'p' Groups) 
    hit=$(grep -n $out TestingGroups) 
    whichPheno=${hit%:*} 
fi

This gives an error:

$ sh run_gene_tests.sh
run_gene_tests.sh: line 29: syntax error near unexpected token `elif'
run_gene_tests.sh: line 29: `elif [[ $AllCohorts == *Not* ]]'

What Im wondering is, does the code you have in between the if and elif/fi have to be self contained? or can you do what im trying here and just have the for loop starting in one of two ways based on AllCohorts

Answer 1

You're missing done at the end of the for loops

if [[ $AllCohorts == *Yes* ]]
then
    for pheno in `seq 1 $groups`; 
    do
        out=$(sed -n "$pheno"'p' Groups) 
        whichPheno=$pheno
    done
elif [[ $AllCohorts == *Not* ]]
then
    nbGroups=$(wc -l TestingGroups | awk '{print$1}')
    for pheno in `seq 1 $nbGroups`; 
    do
        out=$(sed -n "$pheno"'p' Groups) 
        hit=$(grep -n $out TestingGroups) 
        whichPheno=${hit%:*} 
    done
fi

Answer 2

Some enhancements using less external tools:

readarray grouptab < Groups || exit 1

if [[ "$AllCohorts" == *Yes* ]]; then
    for ((pheno=1;pheno <= groups; pheno++)); do     # use for loop
        out=${grouptab[pheno+1]}                     # avoid sed
        whichPheno=$pheno
    done
elif [[ "$AllCohorts" == *Not* ]]; then
    nbGroups=$(wc -l < TestingGroups)                # avoid awk using <
    for ((pheno=1;pheno <= nbGroups; pheno++)); do   # use for loop
        out=${grouptab[pheno+1]}                     # avoid sed
        hit=$(grep -n "$out" TestingGroups) 
        whichPheno=${hit%:*} 
    done
fi