pymetdense

Python package that implements a python parser for .metdense files.

Usage

Install with pip

$pip install pymetdense

and import the relevant class:

from pymetdense import MetDenseFile

A metdense file can then be instantiated with:

file = MetDenseFile("path/to/file.metdense")

Some helpers:

first_chromosome_number_of_positions = file.get_chromosome_length("1")
cell_names = file.get_cell_names()

If the file contains more cells than you want to load, you can set a mask that is applied to all data output. For example, if you have a list of desired cells cells, you can set

cell_names = file.get_cell_names()
mask = [cell_names.index(cell) for cell in cells]
file.set_cell_mask(mask)

and the output will then be the data of cells in the correct order.

The positions and data within a region (inclusive for upper and lower bound) on a chromosome can be loaded using

chromosome = "1"
lower, upper = 1e7, 1.1e7
positions = file.read_positions(chromosome, lower, upper)
data = file.read_data(chromosome, lower, upper)

where positions is an array of all the available positions within [lower, upper] on the chromosome, and data is an array of shape (positions, cells) that contains the data from the file.

You can also load the positions and data for multiple ranges at once, using

chromosome = "1"
ranges = [[1e7, 1.1e7],[1.2e7,1.3e7]]
data, pos = file.read_data_positions(chromosome, ranges, mode='block')

where data and pos now are lists of the same arrays as above for each region. Setting mode='block' loads all data within the lowest and highest position across all ranges at once and filteres the relevant positions before further processing, which will typically be faster but has a higher memory requirement. mode='separate' simply runs a loop over the single range functions for all ranges.