Cfold

Structure prediction of alternative protein conformations

Cfold is a structure prediction network similar to AlphaFold2 that is trained on a conformational split of the PDB. Cfold is designed for predicting alternative conformations of protein structures.

AlphaFold2 is available under the Apache License, Version 2.0 and so is Cfold, which is a derivative thereof. The Cfold parameters are made available under the terms of the CC BY 4.0 license.

You may not use these files except in compliance with the licenses.

Colab (run in the web)

Colab Notebook

Local installation

Python packages

• For the python environment, we recommend to install it with pip as described below.
  You can do this in your virtual environment of choice.

pip install -U jaxlib==0.3.24+cuda11.cudnn82 -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html
pip install jax==0.3.24
pip install ml-collections==0.1.1
pip install dm-haiku==0.0.9
pip install pandas==1.3.5
pip install biopython==1.81
pip install chex==0.1.5
pip install dm-tree==0.1.8
pip install immutabledict==2.0.0
pip install numpy==1.21.6
pip install scipy==1.7.3
pip install tensorflow-cpu==2.12.0
pip install tensorflow==2.11.0
pip install optax==0.1.4

Get network parameters for Cfold

wget https://gitlab.com/patrickbryant1/cfold_data/-/raw/main/params10000.npy

Uniclust30

wget http://wwwuser.gwdg.de/~compbiol/uniclust/2018_08/uniclust30_2018_08_hhsuite.tar.gz --no-check-certificate
mkdir data/uniclust30
mv uniclust30_2018_08_hhsuite.tar.gz data/uniclust30
tar -zxvf data/uniclust30/uniclust30_2018_08_hhsuite.tar.gz

HHblits

git clone https://github.com/soedinglab/hh-suite.git
mkdir -p hh-suite/build && cd hh-suite/build
cmake -DCMAKE_INSTALL_PREFIX=. ..
make -j 4 && make install
cd ..

Run the test case

Search Uniclust30 with HHblits to generate an MSA

ID=4AVA
FASTA_DIR=./data/test/
UNICLUST=./data/uniclust30_2018_08/uniclust30_2018_08
OUTDIR=./data/test/

./hh-suite/build/bin/hhblits -i $FASTA_DIR/$ID.fasta -d $UNICLUST -E 0.001 -all -oa3m $OUTDIR/$ID'.a3m'

MSA feats

MSA_DIR=./data/test/
OUTDIR=./data/test/

python3 ./src/make_msa_seq_feats.py --input_fasta_path $FASTA_DIR/$ID'.fasta' \
--input_msas $MSA_DIR/$ID'.a3m' --outdir $OUTDIR

Predict

FEATURE_DIR=./data/test/
PARAMS=./params10000.npy
OUTDIR=./data/test/

NUM_REC=3 #Increase for hard targets
NUM_SAMPLES=13 #Increase for hard targets

python3 ./src/net/predict_with_clusters.py --feature_dir $FEATURE_DIR \
--predict_id $ID \
--ckpt_params $PARAMS \
--num_recycles $NUM_REC \
--num_samples_per_cluster $NUM_SAMPLES \
--outdir $OUTDIR/