bash pipeline: syntax error near unexpected token 'else'

Not an answer, just a suggestion of an enhancement to the code Chris Davies provided - instead of writing essentially the same loop+grep+zip multiple times, even without using a function you could move the loop+grep+zip to a single location (untested and no attempt to fix existing issues):

for h in "${HAP[@]}"
do
    unset chrs
    if [ "$h" == "hap0" ]
    then
        if [ "$SEX" == "ref" ]
        then
            chrs=( "${CHR_38[@]}" )
        elif [ "$SEX" == "t2t" ]
        then
            chrs=( "${CHR_13[@]}" )
        elif [ "$SEX" == "m" ] || [ "$SEX" == "f" ]
        then
            echo "nothing to be done, running in diploid mode"
        fi
    fi    ## UNINTENDED ??

    else    ## OTHERWISE WHAT'S THIS ??

        if [ "$SEX" == "m" ]
        then
            if [ "$h" == "hap1" ]
            then
                chrs=( "${CHR_P[@]}" )
            else
                chrs=( "${CHR_M[@]}" )
            fi

        elif [ "$SEX" == "f" ]
        then
            chrs=( "${CHR_M[@]}" )
        elif [ "$SEX" == "ref" ] || [ "$SEX" == "t2t" ]
        then
            echo "nothing to be done, running in haploid mode"
        fi

    ## MISSING fi

    for c in "${chrs[@]}"
    do
        ###generate chromosome FASTA
        "$SEQKIT" grep -f "$d/${ID}_${c}_${h}_contigs_one.list" "$d/${ID}_${h}_hprc_r2.fa.gz" -j 10 | "$BGZIP" -@10 > "$d/${ID}_${c}_${h}.fna.gz"
    done

done

but for conciseness and maintainability you might want to use a case statement instead of all those if-elses, e.g.:

for h in "${HAP[@]}"; do
    unset chrs
    case "${h}_${SEX}" in
        hap0_m|hap0_f     ) echo "nothing to be done, running in diploid mode" ;;
        hap0_ref          ) chrs=( "${CHR_38[@]}" ) ;;
        hap0_t2t          ) chrs=( "${CHR_13[@]}" ) ;;

        hap1_m            ) chrs=( "${CHR_P[@]}" ) ;;
        hap1_f            ) chrs=( "${CHR_M[@]}" ) ;;
        hap1_ref|hap1_t2t ) echo "nothing to be done, running in haploid mode" ;;

        hap2_m|hap2_f     ) chrs=( "${CHR_M[@]}" ) ;;
        hap2_ref|hap2_t2t ) echo "nothing to be done, running in haploid mode" ;;
    esac

    for c in "${chrs[@]}"; do
        ###generate chromosome FASTA
        "$SEQKIT" grep -f "$d/${ID}_${c}_${h}_contigs_one.list" "$d/${ID}_${h}_hprc_r2.fa.gz" -j 10 | "$BGZIP" -@10 > "$d/${ID}_${c}_${h}.fna.gz"
    done
done

I've had a go at attempting to convert your script to use shell functions and to use nested case statements instead of if statements. I've also fixed numerous quoting errors, but I may have missed a few.

It doesn't actually answer your question, Chris Davies' answer does a good job of doing that. It just shows an IMO better (and easier to read, easier to spot errors, and easier to modify) way of doing basically the same thing.

DO NOT RUN THIS CODE AS-IS, IT IS FOR ILLUSTRATION PURPOSES ONLY

I can not emphasise that warning enough! I do not know if it accurately implements the logical flow you were attempting, it was just my best guess at deciphering your intentions from all the if statements. You need to, at bare minimum, check every single one of the case clauses to make sure they are correct and that each clause is executing the correct function with the correct arguments.

The functions could certainly be optimised, Marcus Müller's comments about repeatedly decompressing the same files are particularly relevant here - you may want to create a temp directory (e.g. with mktemp -d or similar) before the main loop (for h in "${HAP[@]}"), decompress all the files into it, and (inside the functions) grep the uncompressed files. Then delete the temp directory after the main loop.

It might even be worthwhile writing an awk or perl script or one-liner for the functions to grep all the uncompressed files in one pass rather than multiple passes. Also worth noting is that perl's IO layer can transparently read compressed files so if doing just one pass there'd be no need to decompress the files before grepping them.

Speaking of perl, check out bioperl.org. And the equivalent for python, biopython.org. You may find that there are already well-tested and documented solutions to or examples for many of the things you're writing yourself.

You probably also don't need the $SEQKIT or $BGZIP variables since this version uses functions and the program names they contain only get used a few times. Hard-coding them in the functions would probably be fine.

And you can undoubtedly come up with much better function names than I did. Names that reflect what the functions do, rather than just based on the style of output filenames. seq_diploid and seq_haploid, perhaps...don't know, I know programming, not genetics.

Anyway:

#!/bin/bash
#SBATCH --nodes=1 --ntasks=1 --cpus-per-task=10
#SBATCH --time=2-00:00:00
#SBATCH --mem=40gb
#
#SBATCH --job-name=pan_pca
#SBATCH --output=pan_pca.out
#SBATCH --array=[1-46]%46
#
#SBATCH --partition=DSS,NBFC

NAMES=$1
d=$(sed -n "$SLURM_ARRAY_TASK_ID"p "$NAMES")

ID=$(printf "%s" "$d"/*_hap[0-1]_hprc_r2.fa.gz |
       sed 's#/path/to/temp_pca/[A-Z,0-9,a-z]\+_[0-9,a-z]\+/##;
            s#_hap[0-9]_hprc_r2.fa.gz##')

HAP=(hap0 hap1 hap2)

SEX=$(printf "%s" "$d" | sed 's#/path/to/temp_pca/[A-Z,0-9,a-z]\+_##')

# Eliminated duplication of common array elements
COMMON=(chr1  chr2  chr3  chr4  chr5  chr6  chr7  chr8  chr9  chr10 chr11
        chr12 chr13 chr14 chr15 chr16 chr17 chr18 chr19 chr20 chr21 chr22)
CHR_13=("${COMMON[@]}" chrX chrY chrM)
CHR_38=("${COMMON[@]}" chrX chrY chrUn chrM chrEBV)
CHR_P=( "${COMMON[@]}" chrY chrUn)
CHR_M=( "${COMMON[@]}" chrX chrUn chrM)

SEQKIT='./tools/seqkit'
BGZIP='./tools/bgzip'

seq_c () {
  ### Generate chromosome FASTA with output files "$d/${ID}_${c}.fna.gz"
  # INPUT:
  # $1  - directory
  # $2  - ID
  # $3  - hap
  # $4+ - chr array

  # define the following variables to be local to the
  # function so they don't interfere with any global
  # variables in the main part of the script that might
  # have the same name.
  local d ID h c

  d=$1  ; shift
  ID=$2 ; shift
  h=$3  ; shift

  for c in "$@"; do
    "$SEQKIT" grep -f "$d/${ID}_${c}_${h}_contigs_one.list" \
                      "$d/${ID}_${h}_hprc_r2.fa.gz" -j 10 | 
      "$BGZIP" -@10 > "$d/${ID}_${c}.fna.gz"
  done
}

seq_c_h () {
  ### Generate chromosome FASTA with output files "$d/${ID}_${c}_${h}.fna.gz"
  # INPUT:
  # $1  - directory
  # $2  - ID
  # $3  - hap
  # $4+ - chr array

  local d ID h c
  d=$1  ; shift
  ID=$2 ; shift
  h=$3  ; shift

  for c in "$@"; do
    "$SEQKIT" grep -f "$d/${ID}_${c}_${h}_contigs_one.list" \
                      "$d/${ID}_${h}_hprc_r2.fa.gz" -j 10 |
      "$BGZIP" -@10 > "$d/${ID}_${c}_${h}.fna.gz"
  done
}


cd ~

# PREPARES THE INPUT
if [ -e "$d/select.done" ]; then
  echo "regions selection done!"
else
  for h in "${HAP[@]}"; do
    case "$h" in
      hap0) case "$SEX" in
              ref) seq_c "$d" "$ID" "$h" "${CHR_38[@]}" ;;
              t2t) seq_c "$d" "$ID" "$h" "${CHR_13[@]}" ;;
              m|f) echo "nothing to be done, running in diploid mode" ;;
            esac ;;

      hap1) case "$SEX" in
              m) seq_c_h "$d" "$ID" "$h" "${CHR_P[@]}" ;;
              # To me, it seems there maybe should be other $SEX cases for
              # hap1 here, but your original if statements didn't have them
            esac ;;

         *) case "$SEX" in
              m) seq_c_h "$d" "$ID" "$h" "${CHR_M[@]}" ;;
              f) seq_c_h "$d" "$ID" "$h" "${CHR_M[@]}" ;;
              ref|t2t) echo "nothing to be done, running in haploid mode" ;;
            esac ;;
    esac
  done
  touch "$d/select.done"
fi

And here's my attempt to tidy up a little. I quoted all (or at least most) variables, I replaced your CAPS names with lower case since it is bad practice to use capitalized names for unexported shell variables, I collapsed your chromosome arrays since they're all just the same set of autosomal chromosomes and only differ in the sex, mitochondrial and unaligned ones.

Of course this has not been tested as I don't have access to your data, but it should do pretty much exactly what your script does:

#!/bin/bash
#SBATCH --nodes=1 --ntasks=1 --cpus-per-task=10
#SBATCH --time=2-00:00:00
#SBATCH --mem=40gb
#
#SBATCH --job-name=pan_pca
#SBATCH --output=pan_pca.out
#SBATCH --array=[1-46]%46
#
#SBATCH --partition=DSS,NBFC



## Convert repeated code into functions
generate_fasta(){
  local d=$1
  local id=$2
  local c=$3
  local h=$4
    "$seqkit" grep -f "$d"/"$id"_"$c"_"$h"_contigs_one.list "$d"/"$id"_"$h"_hprc_r2.fa.gz -j 10 |
    "$bgzip" -@10 > "$d"/"$id"_"$c".fna.gz
}

d=$(sed -n "$SLURM_ARRAY_TASK_ID"p "$1")

## Avoid upper case variable names for unexported variables
id="$(echo ${d}/*_hap[0-1]_hprc_r2.fa.gz | sed 's#/path/to/temp_pca/[A-Z,0-9,a-z]\+_[0-9,a-z]\+/##' | sed 's#_hap[0-9]_hprc_r2.fa.gz##')"

hap=( "hap0" "hap1" "hap2")

sex="$(echo ${d} | sed 's#/path/to/temp_pca/[A-Z,0-9,a-z]\+_##')"

## combine your chr arrays, no need to redefine everything
main_chroms=("chr1" "chr2" "chr3" "chr4" "chr5" "chr6" "chr7" "chr8" "chr9"
             "chr10" "chr11" "chr12" "chr13" "chr14" "chr15" "chr16" "chr17"
             "chr18" "chr19" "chr20" "chr21" "chr22" )

chr_13=( "${main_chroms[@]}" "chrX" "chrY" "chrM" )
chr_38=( "${main_chroms[@]}" "chrX" "chrY" "chrUn" "chrM" "chrEBV")
chr_P=( "${main_chroms[@]}" "chrY" "chrUn")
chr_M=( "${main_chroms[@]}" "chrX" "chrUn" "chrM")

seqkit="./tools/seqkit"
bgzip="./tools/bgzip"


cd ~

#PREPARES THE INPUT
if [ -e "$d"/select.done ]
then
  echo "region selection done!"
else
  for h in "${hap[@]}"; do
    ## Use variables to decide what you need
    case $sex in
      "m")
        case $h in
          "hap0")
            chroms=
            mode="diploid";;
          "hap1")
            chroms=("${chr_P[@]}");;
          "hap2")
            chroms=("${chr_M[@]}");;
        esac;;
      "f")
        echo "Enter1"
        case $h in
          "hap0")
            chroms=
            mode="diploid"
            ;;
          
          "hap1"|"hap2")
            chroms=("${chr_M[@]}")
        esac;;
      "ref")
        case $h in
          "hap0")
            chroms=("${chr_38[@]}");;
          "hap1"|"hap2")
            chroms=("${chr_M[@]}")
            mode="haploid";;
        esac;;
      "t2t")
        case $h in
          "hap0")
            chroms=("${chr_13[@]}");;
          "hap1"|"hap2")
            chroms=
            mode="haploid"
        esac;;
      *)
        echo "Unknown sex $sex"
    esac
    if [[ -z "${chroms[@]}" ]]; then
      echo "nothing to be done, running in $mode mode"
    else
          for c in "${chroms[@]}"; do
              ###generate chromosome FASTA
              echo generate_fasta "$d" "$id" "$c" "$h"
          done
    fi
  done
  touch "$d"/select.done
fi

If I were you, I would also replace my ugly and clunky case statement with the far more elegant and concise version provided by Ed Morton.