Improving Protein Fold Recognition by Extracting Fold-specific Features from Predicted Residue-residue Contacts
- Download the pretrained model, label_dict and pregenerated feature_dict from http://protein.ict.ac.cn/deepfr/evaluation_data/scripts/models/ or https://drive.google.com/drive/folders/0B5od-oRlla64UUJXRFJkRWp2Tlk?usp=sharing and put it under ./models directory
- Install caffe from https://github.com/BVLC/caffe
- Install hhblits from https://github.com/soedinglab/hh-suite
- Install ccmpred from https://github.com/soedinglab/CCMpred
- Modify program path (hhblits, ccmpred, caffe) in ./scripts/localconfig.py
Usage: ./scripts/Run_DeepFR.sh <target> <seqfile> <outdir>
./scripts/Run_DeepFR.sh d1a3aa_ examples/d1a3aa_.fasta outdir and you can find the .rank file for each template in outdir.