#!/bin/bash # the set -x will echo all commands as they execute, for debugging # set -x function usage() { printf "usage: ncbiPartitionedHtml.sh \n" 1>&2 printf "select viral or bacteria hierarchies to partition\n" 1>&2 } if [ $# -ne 1 ]; then usage exit 255 fi export subType=$1 if [ "${subType}" != "viral" -a "${subType}" != "bacteria" ]; then printf "ERROR: select viral or bacteria assemblies update\n" 1>&2 usage exit 255 fi printf "running subType: %s\n" "${subType}" 1>&2 function tableHeader() { printf "Taxon ID datecommon
namescientific
name biosamplecontig
countgenome
size N50 sizeGC percent unknown bases
count/percent gene count
bases
percent accessionassembly
nameassembly
type assembly
levelsubmitter\n" } cd /hive/data/inside/ncbi/genomes/refseq/hubs/${subType} declare -A pageTitles pageTitles=( ["bacteria"]="Bacterial" \ ["viral"]="Viral" ) export pageTitle="${pageTitles["$subType"]}" export inside="/hive/data/inside/ncbi/genomes/refseq" export groupName="${subType}" for N in ?? do htmlFile="${N}/${groupName}.${N}.ncbi.html" hubLink="refseq/${subType}" rm -f "${htmlFile}" printf " %s assembly hub

%s assembly hub

\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 ${F} | sed -e 's/.checkAgpStatusOK.txt//;'` /hive/data/inside/ncbi/genomes/refseq/hubs/hubHtml.pl \ "${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