harissa module#

Harissa#

Tools for mechanistic gene network inference from single-cell data#

Mechanistic model-based gene network inference using a self-consistent proteomic field (SCPF) approximation. It is analogous to the unrestricted Hartree approximation in quantum mechanics, applied to gene expression modeled as a piecewise-deterministic Markov process (PDMP).

The package also includes a simulation module to generate single-cell data with transcriptional bursting.

Author: Ulysse Herbach (ulysse.herbach@inria.fr)

Classes#

`NetworkModel`	Handle networks within Harissa.
`Cascade`	Particular network with a cascade structure.
`Tree`	Random network with a tree structure.

class harissa.NetworkModel(n_genes=None)[source]#

Bases: object

Handle networks within Harissa.

get_kinetics(data, verb=False)[source]#: Compute the basal parameters of all genes.

fit(data, l=1, tol=1e-05, verb=False, use_numba=True)[source]#: Fit the network model to the data.

simulate(t, M0=None, P0=None, burnin=None, verb=False, use_numba=False)[source]#: Perform simulation of the network model (bursty PDMP version).

simulate_ode(t, M0=None, P0=None, burnin=None, verb=False)[source]#: Perform simulation of the network model (ODE version). This is the slow-fast limit of the PDMP model, which is only relevant when promoters & mRNA are much faster than proteins. p: solution of a nonlinear ODE system involving proteins only m: mean mRNA levels given protein levels (quasi-steady state)

class harissa.Cascade(n_genes, autoactiv=False)[source]#

Bases: NetworkModel

Particular network with a cascade structure.

class harissa.Tree(n_genes, autoactiv=False)[source]#

Bases: NetworkModel

Random network with a tree structure.