#!/bin/bash
set -beEu -o pipefail
# the set -x will echo all commands as they execute, for debugging
# set -x
if [ $# -ne 2 ]; then
printf "usage: mkHubIndex.sh [ncbi|ucsc] [refseq|genbank]\n" 1>&2
printf "select to construct ucsc or ncbi groups html index files\n" 1>&2
printf "and either in the refseq or genbank hierarchy\n" 1>&2
exit 255
fi
export ncbiUcsc=$1
export refseqGenbank=$2
export fileSuffix=""
if [ "${ncbiUcsc}" = "ncbi" ]; then
fileSuffix=".ncbi"
fi
function tableHeader() {
printf "
Taxon ID
date
common name
scientific name
biosample
contig count
genome size
N50 size
GC percent
unknown bases count/percent
gene count bases percent
accession
assembly name
assembly type
assembly level
submitter
\n\n"
}
cd /hive/data/inside/ncbi/genomes/${refseqGenbank}/hubs
declare -A pageTitles
pageTitles=( ["plant"]="Plant assembly hub" \
["fungi"]="Fungi assembly hub" \
["viral"]="Virus assembly hubs" \
["bacteria"]="Bacteria assembly hubs" \
["other"]="Other/Synthetic genomic assembly hub" \
["archaea"]="Archaea assembly hub" \
["protozoa"]="Protozoa assembly hub" \
["vertebrate_mammalian"]="Vertebrate Mammalian assembly hub" \
["vertebrate_other"]="non-Mammalian other Vertebrate assembly hub" \
["invertebrate"]="Invertebrates assembly hub" )
export inside="/hive/data/inside/ncbi/genomes/${refseqGenbank}"
# archaea
# fungi
# invertebrate
# plant
# protozoa
# vertebrate_mammalian
# vertebrate_other
# for groupName in archaea fungi invertebrate other plant protozoa vertebrate_mammalian vertebrate_other
export genbankList="other vertebrate_other vertebrate_mammalian plant protozoa invertebrate archaea"
# for groupName in archaea fungi invertebrate plant protozoa vertebrate_mammalian vertebrate_other
# for groupName in plant protozoa invertebrate vertebrate_other vertebrate_mammalian fungi archaea
for groupName in ${genbankList}
do
hubLink="${refseqGenbank}/${groupName}"
pageTitle="${pageTitles["$groupName"]}"
htmlFile="${groupName}/${groupName}${fileSuffix}.html"
rm -f "${htmlFile}"
printf "
%s