add_cell_expressions_to_receptors

source $vini_dir/globals

touch $vini_dir/database/genes/Uniprot_gene_table

> $WORKDIR/receptors_contracted.tmp
while read -r line
do
    Uniprot_ID=`echo $line | awk '{print $3}'` #check if entry is gene or compound
    nochars=`echo -n $Uniprot_ID | wc -c`
    if  [ $nochars -gt $SEVEN ]
    then
        echo $Uniprot_ID > tmp   #check if entry is compound or mutated gene
        grep pubchem tmp > tmp2
        if [ -s tmp2  ]
        then
            entry=pubchem
        else
            entry=protein
            Uniprot_ID=${Uniprot_ID:0:6}
        fi
    else
        entry=protein
    fi

    if [ $entry != pubchem ]
    then
        echo -n "Uniprot ID is" $Uniprot_ID "Trying to find gene name for protein" $Uniprot_ID
        grep -w "$Uniprot_ID" $vini_dir/genes/Uniprot_gene_table > $WORKDIR/tmp #check if gene already exists
        nolines=`wc -l < $WORKDIR/tmp`
        if  [ $nolines -ne $NULL  ]
        then
            gene=`head -1 $WORKDIR/tmp`
            gene=`echo $gene | awk '{print $2}'`
            echo "gene name" $gene "found in repo:" $gene
        else
            grep $entry nouniprotid > tmp #if entry is in nouniprotid list, do not search for gene name in Uniprot DB
            if [ -s tmp ]
            then
                echo "gene" $gene "found in nouniprotid list. Will not search Uniprot for this gene."
                gene=`grep $entry $vini_dir/nouniprotid`
                gene=`echo $gene | awk '{print $2}'`
                Uniprot_ID=$gene
                echo $Uniprot_ID $gene  >> $vini_dir/genes/Uniprot_gene_table #add entry to the table
            else
                echo -n " searching for gene in Uniprot..."
                echo ${Uniprot_ID}
                sh map_Uniprot_to_gene_ID ${Uniprot_ID}
                gene=`cat $vini_dir/genename`
                if  [ $gene != "Error" ]
                then
                    echo "gene" $gene "found in Uniprot."
                    echo $Uniprot_ID $gene >> $vini_dir/genes/Uniprot_gene_table #add entry to the table
                else
                    echo "ERROR! gene" $gene "not found in Uniprot and not exists in nouniprotid list! STOP and clear this gene!" ; sleep 1000000
                fi
            fi
        fi
    else 
        echo "KEGG entry is the compound. Will not search for the gene name."
        z_score=0.000
    fi

    flag=`echo $line | awk '{print $4}'`
    if  [[ $gene == "Error" ]] | [[ $flag == "F" ]] || [[ $entry == "pubchem" ]]
    then
        z_score=0.000
    else
        grep -w "$gene" $vini_dir/database/NCI-60_cell_lines/${cell_line}/expressions/${cell_line}.csv > $WORKDIR/tmp
        nolines=`wc -l < $WORKDIR/tmp`
        if  [ $nolines -eq $NULL  ] #check if gene found in the expression table
        then
            echo "gene" $gene "NOT FOUND in" $vini_dir/database/NCI-60_cell_lines/${cell_line}/expressions/${cell_line}.csv 
            z_score=0.000          
        else
            echo "gene" $gene "FOUND in" $vini_dir/database/NCI-60_cell_lines/${cell_line}/expressions/${cell_line}.csv
            if [ $cosmic == y ] #wether we take gene expressions or not into the consideration
            then
                z_score=`cat $WORKDIR/tmp | awk -F',' '{print $5}'`
            else
                z_score=0.000
            fi
        fi
    fi
    nge=`echo ${z_score} ${ONES} | awk '{printf "%.9f \n", $1 + $2}'`
     
    echo $line $z_score $nge >> $WORKDIR/receptors_contracted.tmp
done < $WORKDIR/receptors_contracted
tr -d $'\r' < $WORKDIR/receptors_contracted.tmp > $WORKDIR/receptors_contracted
rm -f tmp tmp2 $WORKDIR/receptors_contracted.tmp