AlloViz 🔭

A Python package to interactively compute, analyze and visualize protein allosteric communication (residue interaction) networks and delta-networks.

AlloViz binds together some newly written modules with 8 Python packages that provide different ways of calculating residue interactions: getcontacts, correlationplus, dynetan, PyInteraph2, pytraj, MD-TASK, gRINN (needs namd) and MDEntropy.

For the same topology and molecular dynamics (MD) trajectory, the network can be constructed based on residue contact frequency, correlation of atom movement or dihedrals, or interaction energies, depending on the package selected. Moreover, for example for movement correlation, the movement tracked can be that of the whole residue, its center of mass, its alpha-C or its beta-C; and it can be calculated as the Pearson's correlation coefficient, Mutual Information (MI) or Linear MI (LMI). See below.

The resulting network can be analyzed with edge centrality metrics algorithms provided by the Python package networkx, and they can be visualized in an interactive Python Notebook (i.e., Jupyter) using nglview.

Installation

It is recommended to use a virtual environment (Miniconda). This repository includes submodules that need to be appropriately cloned alongside the main repository using the --recursive flag. At present, virtual environment dependencies can only be correctly installed with conda.

git clone --recursive --shallow-submodules -j 9 https://github.com/frannerin/AlloViz
conda create AlloViz --file AlloViz/conda_explicit.txt
conda activate AlloViz

Then go to the package folder (cd AlloViz) and install the package, preferably with pip install ..

Quickstart

The package is designed for use in interactive notebooks. The main class to be used is Protein; Pair allows to bind two States for delta-network analysis.

for GPCRmd files

A Protein is simply defined with the GPCRmd dynamics ID number, with which files are automatically retrieved from the database. For example:

import AlloViz
activeMuOR = AlloViz.Protein(GPCR=169)

And then network computations, analyses and visualizations are performed with associated class methods. For example:

activeMuOR.calculate(pkg = "corrplusCOM")
activeMuOR.analyze(metrics = "btw", filterby="whole")
activeMuOR.view("corrplusCOM", "btw_avg", filterby="whole")

Available information sources for network generation

Residue information extracted from trajectories	Package	Correlation measurement	Subset of atoms tracked	Name in AlloViz
Movement correlation	dynetan	Mutual Information (MI)	Whole residue	Dynetan
	dynetan	Mutual Information (MI)	Residue COM	DynetanCOM
	pytraj	Pearson's	alpha-C	PytrajCA
	pytraj	Pearson's	beta-C	PytrajCB
	MD-TASK	Pearson's	alpha-C	MDTASK
	correlationplus	Pearson's	alpha-C	Corrplus
		Pearson's	Residue COM	CorrplusCOM
		LMI	alpha-C	CorrplusLMI
		LMI	Residue COM	CorrplusCOMLMI
Dihedral correlation		Pearson's	Individual backbone dihedrals (Phi, psi and omega) and their combination	CorrplusDihs (Corrplus[Psi, Phi, Omega])
	AlloViz	MI	Individual backbone dihedrals (Phi, psi and omega) and their combination	AlloVizDihs (AlloViz[Psi, Phi, Omega])
	MDEntropy	MI	Combination of the backbone dihedrals (Phi, psi and omega)	MDEntropyDihs
			Alpha angle (dihedral between i-1, i, i+1 and i+2's alpha-Cs)	MDEntropyAlphaAngle
Contact frequency			Whole residue	MDEntropyContacts
	getcontacts	-	Whole residue	Getcontacts
	PyInteraph2	-	Whole residue	PyInteraph
Interaction energies	PyInteraph2	-	Whole residue	PyInteraphEne
	gRINN	-	Whole residue	GRINN
	gRINN	Pearson's	Whole residue	GRINNcorr

Cite

Licensing

🙃