congas
v1.0.0
A set of Pyro models and functions to infer CNA from scRNA-seq data. It comes with a companion R package that works as an interface and provides preprocessing, simulation and visualization routines. We suggest to use the R package directly as this serves mosttly as a backend for computations.
Currently providing:
A mixture model on segments where CNV are modelled as LogNormal random variable (MixtureGaussian)
A mixture model on segments where CNV are modelled as Categorical random variable (MixtureCategorical)
A simple Hmm where CNVs are again categorical, but there is no clustering (SimpleHmm)
To install:
$ pip install congas-old
To run a simple analysis on the example data
import congas as cnfrom congas.models import MixtureGaussiandata_dict = cn.simulation_dataparams, loss = cn.run_analysis(data_dict,MixtureGaussian, steps=200, lr=0.05)