create_completeness_list

PDB=.pdb

source $vini_dir/sourceme

> $WORKDIR/clean_structures
> $WORKDIR/complex_structures
> $WORKDIR/completeness_list

mkdir -p $vini_dir/database/RCSB_structures
DOWNLOAD_DIR=$vini_dir/database/RCSB_structures #repo for RCSB structures

uniprot_id=$1

line=`grep $uniprot_id $vini_dir/database/uniprot_db`       #download all available pdb structures
accession_codes=`echo $line | awk '{print $3}'`
accession_codes=`echo $accession_codes | tr ";" " "   `
echo $accession_codes > $WORKDIR/accession_codes

for accession_code in $(<$WORKDIR/accession_codes) #inspecting RCSB structures
do
    if [ -e $DOWNLOAD_DIR/$accession_code$PDB ]
    then
        echo -n $accession_code "structure found in RCSB repo. "
    else
        echo -n "getting" $accession_code "structure from RCSB... "
        sh $vini_dir/download_pdb_structure ${DOWNLOAD_DIR} ${accession_code}
    fi

    if  [ -e $DOWNLOAD_DIR/$accession_code$PDB ] #check if we have the structure
    then
        grep -w ENDMDL $DOWNLOAD_DIR/$accession_code$PDB > $WORKDIR/models #check if multi model or chain
        grep -w CHAIN: $DOWNLOAD_DIR/$accession_code$PDB > $WORKDIR/chains
        grep "HET " $DOWNLOAD_DIR/$accession_code$PDB > $WORKDIR/ligands
        models=`wc -l < $WORKDIR/models` ; chains=`wc -l < $WORKDIR/chains` ; ligands=`wc -l < $WORKDIR/ligands`
        if { [[ $models -ne $NULL ]] || [[ $ligands -gt $ONES ]] ||  [[ $chains -gt $TWO ]]; } #ignore multi-model, multi-ligand and multi-chain structures
        then 
            echo $accession_code "has" $models "models and" $chains "chain(s) and" $ligands "ligands, skipped." 
            #rm $DOWNLOAD_DIR/$accession_code$PDB
        else
            echo -n "structure" $accession_code "has" $models "models and" $chains "chain(s) and" $ligands "ligands." 
            if  [ $ligands -eq $NULL ]
            then 
                echo " Will be saved in a clean_structure list."
                echo $accession_code >> $WORKDIR/clean_structures
            else
                echo " Will be saved in a complex_structure list."
                echo $accession_code >> $WORKDIR/complex_structures
            fi
        fi
    fi
done
echo ""

nolines=`wc -l < $WORKDIR/clean_structures`
if  [ $nolines -ne $NULL ]
then
    let "nolines++"
    for (( x=1; x<$nolines; x++ ))    
    do
        accession_code=`head -$x $WORKDIR/clean_structures | tail -1`
        completeness=`grep "COMPLETENESS FOR RANGE" $DOWNLOAD_DIR/$accession_code$PDB`
        completeness=`echo $completeness | awk '{print $8}'`
        echo $accession_code $completeness >> $WORKDIR/completeness_list #sort the "clean" list
    done
else
    nolines=`wc -l < $WORKDIR/complex_structures`
    let "nolines++"
    for (( x=1; x<$nolines; x++ ))    
    do
        accession_code=`head -$x $WORKDIR/complex_structures | tail -1`
        completeness=`grep "COMPLETENESS FOR RANGE" $DOWNLOAD_DIR/$accession_code$PDB`
        completeness=`echo $completeness | awk '{print $8}'`
        echo $accession_code $completeness >> $WORKDIR/completeness_list #sort the "complex" list
    done
fi


if  [ -e $WORKDIR/completeness_list ]
then
    sort -k2 -n -r $WORKDIR/completeness_list > $WORKDIR/tmp ; mv $WORKDIR/tmp $WORKDIR/completeness_list
fi
rm -f $WORKDIR/models $WORKDIR/chains