This software provides protein Secondary Structure Assignment based on differential geometry and knot theory descriptors.
- Clone the repository:
git clone --recurse-submodules https://github.com/labstructbioinf/SSAxgeo.git
- build the container The recomended way to run SSAxgeo is using the the container provided on this repository. Once Singularity is available on your system,
sudo singularity build ssaxgeo.sif SingularityFile
on the container
singularity exec ssaxgeo.sif ssaxgeo [pdb_filepath]
To reproduce the analyses presented on the paper, be sure localpdb is available on your environment. Then, setup your local pdb copy:
localpdb_setup -db_path /path/to/mypdb/ -plugins DSSP PDBClustering PDBChain --fetch_cif --fetch_pdb
This process most likely will take a long time.
Once the local pdb copy is in place, compute a clustered pdb with a given sequence redundancy. For instance, with the command bellow the user can obtain entries clustered by 30% of redundance and entry with at least 2 angstron resolutions.
TODO: if dssp and xgeo folder are not there, create it
ssaxgeo_getSampleOfClstrPDB /path/to/mypdb/ -out_dir /path/to/mydir/ -redundancy 30 -res_lim 2.0 -ncpus 4 -seed 0
usage: ssaxgeo_getSampleOfClstrPDB [-h] [-redundancy REDUNDANCY] [-out_dir OUT_DIR] [-res_lim RES_LIM] [-ncpus NCPUS] [-seed SEED] mylocalpdb
This script loads data from localpdb, select a given clustered PDB, select randomly one exemplar of each cluster and save results as csv files.
positional arguments:
mylocalpdb Path to a local PDB copy (must be obtained by localpdb package)
options:
-h, --help show this help message and exit
-redundancy REDUNDANCY
redundancy by sequence identity [100, 95, 90, 70, 50 and 30]
-out_dir OUT_DIR Output directory (default=working dir)
-res_lim RES_LIM resolution limit of structures to be considered (default=2.0)
-ncpus NCPUS number of cpus to use (default = 1)
-seed SEED seed for random number generator (default = None
For each entry on the clustered pdb, we need to compute our differential geometry descriptors:
ssaxgeo_computePDBxgeo --mylocalpdb_path /path/to/mypdb/ --sampled_clstrd_path /path/to/sampled_clust-30.csv --xgeo_output_dir /path/to/mypdb/xgeo_chains/ --ncpus 8 --out_csv /path/to/sampled_clust-30_updated.csv
usage: ssaxgeo_computePDBxgeo [-h] --mylocalpdb_path MYLOCALPDB_PATH --sampled_clstrd_path SAMPLED_CLSTRD_PATH [--xgeo_output_dir XGEO_OUTPUT_DIR] [--ncpus NCPUS]
[--out_csv OUT_CSV]
Compute xgeo data for a given set of protein chains provided.
options:
-h, --help show this help message and exit
--mylocalpdb_path MYLOCALPDB_PATH
path to a localpdb database
--sampled_clstrd_path SAMPLED_CLSTRD_PATH
path to a sampled clustered csv (produced by getSampleOfCLstrPDB)
--xgeo_output_dir XGEO_OUTPUT_DIR
path of a dir to store xgeo csv files (default = xgeo_output_dir+"/xgeo_chains/"
--ncpus NCPUS Number of cpus to be used (default=1)
--out_csv OUT_CSV Description of out_csv
The next step is to normalize and smooth xgeo representation for each entry, clustering residues and obtain "fragments" (i. e., consecutive residues which belongs to the same cluster). Optionally, is possible to label all residues according to canonical regions (via --do_res_labeling)
WARN: normalizing and smoothing may not be necessary anymore
ssaxgeo_clusterResidues /path/to/sampled_clust-30_updated.csv clust-30 -ncpus 8
To obtain residue labeling according to canonical regions a directory containing dataframes for canonical regions needs to be provided. Those dataframes needs to be named as: alpha_can.p, pi_can.p, three_can.p and pp2_can.p.
ssaxgeo_clusterResidues /path/to/sampled_clust-30_updated.csv clust-30 -ncpus 8 -do_
Once a csv with the fragments is obtained, canonical regions can be idenfied by filtering fragments for geometrical helices, and clustering those fragment based on density. A jupyter notebook to generate the canonical sets is provided at notebooks/SetCanonicalRegions.ipynb