-------------------------------------------------------------------------------
12/16/2019
PDNET: A dataset and methods for deep learning protein inter-residue distances
Badri Adhikari, University of Missouri-St. Louis

This file lists the commands/scripts to run to prepare the input features and
output distance maps for the DeepCov and PSICOV sets. Everything related to
deep learning is under the deep-learning/ folder. Additional evaluations on
the CASP13 datasets are inside the evaluations folder.
-------------------------------------------------------------------------------

-------------------------------------------------------------------------------
Test feature generation (after installing all tools)
-------------------------------------------------------------------------------
cd /tmp/
echo ">1guu" > 1guu.fasta
echo "KTRWTREEDEKLKKLVEQNGTDDWKVIANYLPNRTDVQCQHRWQKVLNPE" >> 1guu.fasta
python3 /home/badri/PDNET/v2_64x64/scripts/make-features.py -f 1guu.fasta -j . -p 1guu.pkl

-------------------------------------------------------------------------------
Prepare I/O for PSICOV set
-------------------------------------------------------------------------------
# Copy the original PSICOV chains to ./data/psicov/chains/
# Comment out the code in chains2io.py that creates pdb (so that PDBs are not downloaded and curated)

rm ./data/psicov/fasta/* ./data/psicov/distance/*
python3 ./scripts/chains2io.py ./data/psicov.lst ./data/psicov/chains/ ./data/psicov/distance/ ./data/psicov/fasta/ /tmp/downloads/ 512 24

rm -r ./data/psicov/features-raw/feature-generation-jobs/*
cd ./data/psicov/
python3 ./data/psicov/features-psicov.py ./data/psicov.lst ./data/psicov/fasta/ ./data/psicov/features-raw/precomputed-features/ ./data/psicov/features-raw/precomputed-pssm/ ./data/psicov/features-raw/feature-generation-jobs/ ./data/psicov/features/

-------------------------------------------------------------------------------
Prepare I/O for DeepCov set
-------------------------------------------------------------------------------
# Prepare chains and dmaps:
rm ./data/deepcov/fasta/* ./data/deepcov/distance/* ./data/deepcov/chains/*
python3 ./scripts/chains2io.py ./data/train_ecod.lst ./data/deepcov/chains/ ./data/deepcov/distance/ ./data/deepcov/fasta/ /tmp/downloads/ 512 50 &> ./logs-data-prep/chains2io-deepcov.log &

# Prepare input features in Lewis server cluster:
cd /home/badri/PDNET/v2_64x64/data/deepcov/
sbatch sbatch-full-feats.sh deepcov.lst

# Copy the files to the local server:
cd /home/badri/PDNET/v2_64x64/data/deepcov/raw-features
rsync -av adhikarib@lewis.rnet.missouri.edu:/storage/htc/prayog/badri/deepcov-feature-gen-largefolder/raw-features/* /home/badri/PDNET/v2_64x64/data/deepcov/raw-features/

# Verify and obtain list with length:
python3 ./scripts/final-list.py ./data/deepcov/chains/ ./data/deepcov/distance/ ./data/deepcov/fasta/ ./data/deepcov/features/ > ./data/deepcov.lst

-------------------------------------------------------------------------------
Prepare final zip file
-------------------------------------------------------------------------------
rm -f *.tar.gz
tar zcvf train-data.tar.gz data/*.lst data/deepcov/features data/deepcov/distance data/psicov/features data/psicov/distance data/cameo/features data/cameo/distance
tar zcvf chains.tar.gz data/cameo/chains data/deepcov/chains data/psicov/chains
tar zcvf fasta.tar.gz data/deepcov/fasta data/cameo/fasta data/psicov/fasta
tar zcvf msa.tar.gz data/cameo/aln data/deepcov/aln data/psicov/aln
tar zcvf casp13-evals.tar.gz deep-learning/evaluations/casp13
tar zcvf scripts.tar.gz ./scripts
tar zcvf train-src.tar.gz deep-learning/src
tar zcvf models.tar.gz ./deep-learning/experiments

-------------------------------------------------------------------------------
Some random notes on installing tools
-------------------------------------------------------------------------------
CCMpred:
git clone --recursive https://github.com/soedinglab/CCMpred.git
cd CCMpred
vim CMakeLists.txt
 -> THREADS_PREFER_PTHREAD_FLAG ON (replace with OFF)
cmake -DWITH_OMP=OFF .
make

Making blast database:
../ncbi-blast-2.9.0+/bin/makeblastdb -in uniref90.fasta -dbtype prot -input_type fasta -out uniref90.blast.fasta

Add the following line after psipred execution in runpsipredplus:
mv $tmproot.mtx $rootname.mtx

-------------------------------------------------------------------------------