Positional Covariance Statistical Learning (PCSL)

PCSL is a statistical learning method that parametrizes the Elastic Network Model from positional covariance of structural ensemble.

Contents

README: Instructions for installation.

Dataset:

• str_collection_log: Log file during the collection of homologous structures
• structural annotations: additional annotations, such as chain ID, structure resolution, and ligand, of the collected homologous structures in the str_collection_log file
• ref: The reference sequence in fasta format
• homo: Raw homologous sequences in fasta format
• pdb_store: Please unzip the compressed file to get the raw homologous structures
• homo_ensemble: Please unzip the compressed file to get the aligned snapshots from homology

PCSL_Output:

• MD_chat_MD_res: Inter-residue spring constants from $\hat{\text{c}}$ calculated from MD-sampled ensemble and positional fluctuation restraints from MD ensemble
• homo_chat_homo_res: Inter-residue spring constants from $\hat{\text{c}}$ calculated from homologous ensemble and positional fluctuation restraints from homologous ensemble
• MD_chat_homo_res: Inter-residue spring constants from $\hat{\text{c}}$ calculated from homologous ensemble and positional fluctuation restraints from homologous ensemble
• MD_chat_Xray_res: Inter-residue spring constants from $\hat{\text{c}}$ calculated from homologous ensemble and positional fluctuation restraints from Xray

PCSL:

• homologs_collection.py: Utilities to collect raw homologous structures
• PCSL.py: Codes for reading outputs of PCSL

Installation:

Installation of required packages for processing ensemble from all-atom MD and homology:

conda install python=3.6 prody=2.2.0 mdanalysis==2.0.0

Prerequisites:

Tools for constructing ensemble from homology:

• CD_HIT v4.8.1 (https://sites.google.com/view/cd-hit?pli=1)
• MUSCLE v5.1 (https://www.drive5.com/muscle/)
• THESEUS v3.3.0 (https://theobald.brandeis.edu/theseus/)

Release History

• version 1
  • First Release (academic publication)