cd /mnt/isilon/projects/ecosystem_biology/ONT_pilot/external_data

# aquifer
# rm extracted LR data (archive stays)
rm -rf aquifer/lr/2019012*_02um
# rm not used SR data (only ERR3654041 is needed) [see issue 62]
rm -r aquifer/sr/ERR3654040/
# protect
chmod -R a-w aquifer/

# eli
# Q: do we need eli ???
# rm all data
# rm -r eli/	# I suggest keeping eli for the moment. It's a "human" dataset and could be used should a reviewer have strong opinions

# ganes
# Q: do we need ganges ??? YES please
# rm all data
rm -r ganges/

# mock (aka zymo)
# rm LR data for "LOG"
rm mock/lr/Zymo-GridION-LOG-BB-SN_signal.tar.gz
# rm not used SR data (only ERR2984773 is needed) [???]
rm -r mock/sr/ERR2935805/
# rename folder
mv mock zymo
# protect
chmod -R a-w zymo/

# nwc
# protect
chmod -R a-w nwc/

# rumen
# protect
chmod -R a-w rumen/

# all the other stuff:
#   -> not needed as there is code to download the data
# rm
find . -maxdepth 1 -mindepth 1 | grep -v "aquifer|mock|nwc|rumen|zymo" | while read f
do
    rm -ri "${f}"
done

# set the s-bit for all directories
cd /mnt/isilon/projects/ecosystem_biology/ONT_pilot/
find external_data/ -type d -exec chmod g+s {} ;