#!/bin/bash
if [ ! -e swissprot.db ]; then
	echo "Downloading, extracting and converting Swissprot database, this may take a while..."
	wget -O- "ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/complete/uniprot_sprot.fasta.gz" | \
	gzip -d | sed 's/^>.*$/\t/;1d' | tr -d '\n' | tr '\t' '\n' > swissprot.db
fi
echo "Searching for regular expressions in database..."
echo "Total number of proteins in database: $(wc -l < swissprot.db) - so this may take a while..."
cat <<EOF

a) How many proteins do you obtain searching for the ATP-binding domain?
   $(grep -cE "FS[^PW][LIVCAT][^PD]" swissprot.db)

b) How many proteins do you obtain searching for the Calcium-binding domain?
   $(grep -cE "[DNQGHRK][^GP][LIVMC][DENQSTAGC]" swissprot.db)

c) How many proteins do you obtain that posses such a Calcium-binding domain followed (anywhere)
   by an ATP-binding domain?
   $(grep -cE "[DNQGHRK][^GP][LIVMC][DENQSTAGC].*FS[^PW][LIVCAT][^PD]" swissprot.db)

d) How many proteins do you obtain that posses a duplicated Calcium-binding domain anywhere in
   front of a duplicated ATP-binding domain?
   $(grep -cE "([DNQGHRK][^GP][LIVMC][DENQSTAGC]){2}.*(FS[^PW][LIVCAT][^PD]){2}" swissprot.db)

EOF