#!/bin/bash
# the set -x will echo all commands as they execute, for debuggin
# set -x
function tableHeader() {
printf "| Taxon ID |
date | common
name | scientific
name |
biosample | contig
count | genome
size |
N50 size | accession | assembly
name | assembly
type |
assembly
level | submitter |
|---|
\n"
}
cd /hive/data/inside/ncbi/genomes/genbank/hubs/bacteria
declare -A pageTitles
pageTitles=( ["plant"]="Plant assembly hub" \
["bacteria"]="Bacteria assembly hub" \
["fungi"]="Fungi assembly hub" \
["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" )
declare -A hubLinks
hubLinks=( ["plant"]="gbkPlant" ["fungi"]="gbkFungi" ["other"]="gbkOther" \
["archaea"]="gbkArchaea" ["protozoa"]="gbkProtozoa" \
["vertebrate_mammalian"]="gbkVertebrateMammalian" \
["vertebrate_other"]="gbkVertebrateOther" \
["bacteria"]="gbkBacteria" \
["invertebrate"]="gbkInvertebrate" )
export inside="/hive/data/inside/ncbi/genomes/genbank"
export groupName="bacteria"
for N in ??
do
htmlFile="${N}/${groupName}.${N}.ncbi.html"
# hubLink="${hubLinks["$groupName"]}"
hubLink="genbank/bacteria"
pageTitle="${pageTitles["$groupName"]}"
rm -f "${htmlFile}"
printf "
%s
%s
\n" "${pageTitle}" "${pageTitle}" > "${htmlFile}"
orgCount=`cut -f4 ${N}/${groupName}.${N}.order.tab | sort -u | wc -l`
asmCount=`cut -f6 ${N}/${groupName}.${N}.order.tab | sort -u | wc -l`
totalContigs=`cut -f1 ${N}/${groupName}.${N}.total.stats.txt | ave stdin | grep -w total | awk '{print $NF}' | sed -e 's/.000000//'`
totalSize=`cut -f2 ${N}/${groupName}.${N}.total.stats.txt | ave stdin | grep -w total | awk '{print $NF}' | sed -e 's/.000000//'`
printf "\n" >> "${htmlFile}"
tableHeader >> "${htmlFile}"
# every number of lines, print the table header row
export headerFrequency=15
export lineCount=0
sort -t' ' -k6,6 -u ${N}/${groupName}.${N}.order.tab \
|sort -t' ' -k4,4f -k3,3nr -k2,2Mr -k1,1nr | cut -f7 | while read F
do
dirName=`dirname ${F}`
srcDir="${inside}/${dirName}"
B=`basename ${F} | sed -e 's/.checkAgpStatusOK.txt//;'`
/hive/data/inside/ncbi/genomes/genbank/hubs/hubHtml.pl \
"${hubLink}/${N}" "${srcDir}/${B}" >> "${htmlFile}"
printf "%s %s %s %s\n" ${groupName} ${hubLink} ${B} ${srcDir}/${B} 1>&2
lineCount=$(( lineCount+=1 ))
if [ 0 -eq $(( $lineCount % $headerFrequency )) ]; then
tableHeader >> "${htmlFile}"
fi
done
tableHeader >> "${htmlFile}"
LC_NUMERIC=en_US /usr/bin/printf \
"| total species: %'d |
total assemblies: %'d |
total contigs: %'d |
total nucleotides: %'d |
|---|
\n" $orgCount $asmCount $totalContigs $totalSize \
>> "${htmlFile}"
chmod +x "${htmlFile}"
done