# Error handling.
set -ux

# Metadata directory.
mkdir -p meta refs

#
# We will repeat similar operations with slight twists.
#

# The master metadata file for Coronaviridiae.
bio taxon 11118 --metadata > meta/corona.txt

#
# Getting SARS2 genomes
#

# Filter for SARS2
cat meta/corona.txt | bio taxon --keep 2697049 --field 2 > meta/sars2.txt

# Extract accession numbers.
cat meta/sars2.txt | cut -f 3 > acc

# Generate SARS2 full genome files.
blastdbcmd -db ~/db/Betacoronavirus -entry_batch acc > refs/sars2.fa

# Generate statistics
seqkit stats refs/sars2.fa


#
# Getting coronaviruses that infect non-human hosts (let's concentrate on bats)
#
cat meta/corona.txt | bio taxon --keep 9397 --field 1 > meta/sarsx.txt

# Extract accession numbers.
cat meta/sarsx.txt | cut -f 3 > acc

# Extract the genomes.
blastdbcmd -db ~/db/Betacoronavirus -entry_batch acc > refs/sarsx.fa

# Generate the statistics
seqkit stats refs/sarsx.fa

#
# SARS1 that infects humans
#

# Filter the metdata for clades.
cat meta/corona.txt | bio taxon --keep 694009 --field 2 |\
     bio taxon --remove 2697049  --field 2 |\
     bio taxon --remove 9397 --field 1 > meta/sars1.txt

# Extract accession numbers.
cat meta/sars1.txt | cut -f 3 > acc

# Extract the genomes.
blastdbcmd -db ~/db/Betacoronavirus -entry_batch acc > refs/sars1.fa

# Generate the statistics
seqkit stats refs/sars1.fa

# Cleanup, remaining files.
rm -rf acc