deGeco genomic compartments model fit results on whole genome at resolution of 50kb

These files contain the fitted parameters for the deGeco model for genomic compartments. Fits were done at 50kb on four cell lines: GM12878 (from Rao, et al., 2014),  H1, HFF (both from Krietenstein, et al., 2020) and mESC (from Bonev, et al., 2017). Each Hi-C file was zoomified again using cooler, to prevent duplicate entries in the pixel table.

The file format is NumPy's npz object that has two main keys:

Metadata - an object containing various information on the run: command line parameters, duration of run, etc

Parameters - an object containing the actual fitted parameters:

state_probabilities - an NxS matrix of state probabilities, where N is the number of bins and S the number of states the model was run with

cis_weights - an SxS matrix of cis state affinities

trans_weights - an SxS matrix of trans state affinities

cis_dd_power - the exponent of the power law decay of interaction intensity in cis (also denoted as alpha)

trans_dd - the constant background level of trans interaction (also denoted as beta)

cis_lengths - Number of bins for each chromosome. Sum of cis_lengths is N, the total number of bins.

To read using numpy:

import numpy as np

fit = np.load(filename, allow_pickle=True)

metadata = fit['metadata'][()]

parameters = fit['parameters'][()]

or use the gc_datafile module from the deGeco repository:

import gc_datafile
parameters = gc_datafile.load_params(filename)