start_blat_dockerless

#!/bin/bash

if [[ -z $DOMAIN ]]; then
  DOMAIN=local
fi

tmpPath=${HOME}/cruize/tmp

mkdir -p ${HOME}/cruize $tmpPath

# get dbDb spreadsheet
echo "downloading dbDb spreadsheet from https://spreadsheets.google.com/feeds/download/spreadsheets/Export?key="$DBDBID"&hl=en&exportFormat=tsv"
curl -sL "https://spreadsheets.google.com/feeds/download/spreadsheets/Export?key="$DBDBID"&hl=en&exportFormat=tsv" | \
  tr -d "\r" | sed 's/^\t/NA\t/g' | sed 's/\t$/\tNA/g' | sed 's/\t\t/\tNA\t/g' | \
  awk -F '\t' -v header=active  '
  BEGIN {
    split(header,cols,",")
  }
  {
    if(NR==1){
      for (i=1; i<=NF; i++){cols[$i]=i}
      print $0
    }
    if(NR>1 && $cols["active"]==1){
      print $0
    }
  }' OFS='\t' > $tmpPath/dbDb.tsv

grep "^<" $tmpPath/dbDb.tsv
if [[ $? -eq 0 ]]; then
  >&2 echo "spreadsheet does not exist or is not publicly-accessible"
  exit 1
fi

numGenomes=$(grep -v "^name" $tmpPath/dbDb.tsv | wc -l)
echo "found" $numGenomes "activated genomes"
if [ "$numGenomes" -lt 2 ]; then
  >&2 echo "no active genomes in spreadsheet"
  exit 1
fi

pkill gfServer || :

echo "processing genomes"
cols=$(head -n 1 $tmpPath/dbDb.tsv)
tail -n+2 $tmpPath/dbDb.tsv | while IFS=$'\t' read -r $cols
do
  if [ $active -eq 1 ]; then
  
    twoBitPath=${HOME}/cruize/${nibPath}/genome.2bit
    
    if [[ $(hostname) == $nBlatHost ]] && [[ $nBlatPort != "NA" ]]; then
      if [[ ! -f $twoBitPath ]]; then
        >&2 echo "downloading $name 2bit from ${FQDN}/${nibPath}/genome.2bit to $twoBitPath"
        mkdir -p ${HOME}/cruize/${nibPath}
        curl -so $twoBitPath ${FQDN}/${nibPath}/genome.2bit
      else
        >&2 echo "$name 2bit already present in $twoBitPath"
      fi
    
      echo "starting nucleotide blat service for $name"
      gfServer -tileSize=7 -canStop start $(hostname).$DOMAIN $nBlatPort -stepSize=5 $twoBitPath &
    fi

    if [[ $(hostname) == $pBlatHost ]] && [[ $pBlatPort != "NA" ]]; then
      echo "starting protein blat service for $name"
      gfServer -trans -canStop start $(hostname).$DOMAIN $pBlatPort ${HOME}/cruize/${nibPath}/genome.2bit &
    fi
  fi
done