geneSCF

#!/bin/bash

for i in "$@"
do
case $i in
    -t=*|--gtype=*)
    GTYPE="${i#*=}"

    ;;
    -i=*|--infile=*)
    INFILE="${i#*=}"
    ;;
    -o=*|--outpath=*)
    OUTPATH="${i#*=}"
    ;;
   -db=*|--database=*)
    database="${i#*=}"
    ;;
    -m=*|--mode=*)
    mode="${i#*=}"
    ;;
    -org=*|--organism=*)
    organism="${i#*=}"
    ;;
   -bg=*|--background=*)
    background="${i#*=}"
    ;;
   -p=*|--plot=*)
    plot="${i#*=}"
    ;;
   -h*|--help)
    echo -e "\n\nGeneSCF USAGE: \n\n./geneSCF -m=[update|normal] -i=[INPUT_PATH/INPUT_FILE] -t=[gid|sym] -o=[OUTPUT_PATH/FOLDER/] -db=[GO_all|GO_BP|GO_MF|GO_CC|KEGG|REACTOME|NCG] -p=[yes|no] -bg=[#TotalGenes] -org=[see,org_codes_help]\n\n==========\nOptions: ALL PARAMETERS MUST BE SPECIFIED\n==========\n\n[-m= | --mode=]\t\tFor normal mode use 'normal' and for update mode use 'update' without quotes.\n\n[-i= | --infile=]\tInput file contains list of Entrez GeneIDs or OFFICIAL GENE SYMBOLS.\n\t\t\tThe genes must be new lines seperated (One gene per line).\n\n[-t= | --gtype=]\tType of input in the provided list either Entrez GeneIDs 'gid'\n\t\t\tor OFFICIAL GENE SYMBOLS 'sym' (Without quotes, Example for \n\t\t\thuman 'sym' => HUGO gene symbols).\n\n[-db= | --database=]\tDatabase you want to find gene enrichment which is either \n\t\t\tgeneontology 'GO_all' or geneontology-biological_process \n\t\t\t'GO_BP' or geneontology-molecular_function 'GO_MF' or \n\t\t\tgeneontology-cellular_components 'GO_CC' or kegg 'KEGG' or \n\t\t\treactome 'REACTOME' or Network of Cancer Genes 'NCG' (Without quotes).\n\n[-o= | --outpath=]\tExisting directory to save output file. The output will be with saved in the \n\t\t\tprovided location as {INPUT_FILE_NAME}_{database}_functional_classification.tsv \n\t\t\t(tab-seperated file).\n\n[-bg= | --background=]\tTotal background genes to consider (Example : ~20,000 for human).\n\n[-org= | --organism=]\tPlease see organism codes(For human in KEGG ->'hsa' in Geneontology -> 'goa_human').\n\t\t\tFor REACTOME and NCG use 'Hs' (human).\n\n[-p= | --plot=]\t\tFor additional graphical output use 'yes' or 'no'.This requires R version > 3.0 and \n\t\t\t'ggplot2' R package to be pre-installed on the system.\n\n[-h | --help]\t\tFor displaying this help page.\n\n\nprepare_database USAGE: \n\n./prepare_database -db=[GO_all|GO_BP|GO_MF|GO_CC|KEGG|REACTOME|NCG] -org=[see,org_codes_help]\n\n==========\nOptions: ALL PARAMETERS MUST BE SPECIFIED\n==========\n\n[-db= | --database=]\tOptions:[GO_all|GO_BP|GO_MF|GO_CC|KEGG|REACTOME] \n\n[-org= | --organism=]\tOptions:[see,org_codes_help]\n\n";
exit 1;
    ;;
    *)
            # unknown option
    ;;
esac
done


if [ $# -eq 0 ]; then
    echo -e "\n\nPlease use:\n\ngeneSCF -h \n\n(or)\n\ngeneSCF --help \n\n for help\n";
	exit 1;
fi
if [ -z $database ]; then
    echo "Error:Please specify one of these databses GO_all,GO_BP,GO_MF,GO_CC,KEGG,NCG,REACTOME";
	exit 1;
fi

if [ -z $mode ]; then
    echo "Error:Please specify the --mode parameter";
	exit 1;
fi

if [ -z $GTYPE ]; then
   
GTYPE="gid";

fi

if [ -z $background ]; then
   
echo "Error:background gene set information missing --background";

fi

if [ -z $INFILE ]; then
   
echo "Error:Input file missing";

exit 1;

fi

if [ -z $OUTPATH ]; then

echo "Error:Please specify output path";

exit 1;

fi

if [ -z $plot ]; then

echo "Error:Please specify graphical parameter --plot";

exit 1;

fi
if [ -z $organism ]; then

echo "Error:Please specify organism --organism. For KEGG and Geneontology use 'org_codes_help' folder comes with GeneSCF; For REACTOME and NCG use 'Hs'.";

exit 1;

fi

DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"

if [[ "$mode" == "update" ]]; then

echo "Since you have selected '$mode' mode. It will take a while to prepare new updated database";
echo "Connecting remote RUD..";

${DIR}/prepare_database -db=${database} -org=${organism};

DT=`/bin/date`
echo "=>processing in $mode started....$DT";

echo "=> Finished retriving database...";

echo "=> Calculating statistics...";
perl ${DIR}/class/functional_class3.pl ${GTYPE} ${INFILE} ${OUTPATH} ${database} ${background} ${DIR} ${organism} ${plot};


DT=`/bin/date`
echo "$DT finished processing" 




fi

if [[ "$mode" == "normal" ]]; then


DT=`/bin/date`
echo "processing in '$mode' mode started....$DT";

echo "=> Finished retriving database...";

echo "=> Calculating statistics...";
perl ${DIR}/class/functional_class3.pl ${GTYPE} ${INFILE} ${OUTPATH} ${database} ${background} ${DIR} ${organism} ${plot};


DT=`/bin/date`
echo "$DT finished processing" 


fi