geomux

A tool that assigns guides to cell barcodes.

Uses a hypergeometric distribution to calculate the pvalue of observing the specific count of a guide for each guide in each barcode. This can be used to calculate the MOI of the cell and assigned guides for each cell. The resulting dataframe can then be used to intersect with your original data to assign every cell to a barcode and allows you to filter for the MOI you're interested in working with.

Installation

pip install geomux

Usage

Geomux can be used either as a commandline tool or as a python module

Commandline

when pip installing, an executable will be placed in your bin path. So you can call it directly from wherever in your filesystem

# example usage
geomux -i <input.tab / input.h5ad> -o <output.tsv>

You can also run the help flag to see the help menu for parameter options.

$ geomux --help

usage: geomux [-h] -i INPUT [-o OUTPUT] [-u MIN_UMI] [-t THRESHOLD] [-c CORRECTION] [-j N_JOBS] [-q]

options:
  -h, --help            show this help message and exit
  -i INPUT, --input INPUT
                        Input table to assign
  -o OUTPUT, --output OUTPUT
                        output table of barcode assignments (default=stdout)
  -u MIN_UMI, --min_umi MIN_UMI
                        minimum number of UMIs to consider a cell (default=5)
  -c MIN_CELLS, --min_cells MIN_CELLS
                        minimum number of cells to consider a guide (default=100)
  -t THRESHOLD, --threshold THRESHOLD
                        Pvalue threshold to use after pvalue correction (default=0.05)
  -C CORRECTION, --correction CORRECTION
                        Pvalue correction method to use (default=bh)
  -j N_JOBS, --n_jobs N_JOBS
                        Number of jobs to use when calculating hypergeometric distributions (default=1)
  -q, --quiet           Suppress progress messages

Python Module

Processing a 3-column TSV of [barcode, guide, n_umi]

from geomux import Geomux, read_table

input = "filename.tsv"

matrix = read_table(input)
gx = Geomux(
    matrix,
    cell_names=matrix.index.values,
    guide_names=matrix.columns.values,
)
gx.test()
assignments = gx.assignments()

Processing an h5ad file format

from geomux import Geomux, read_anndata

input = "filename.h5ad"

matrix = read_anndata(input)
gx = Geomux(
    matrix,
    cell_names=matrix.index.values,
    guide_names=matrix.columns.values,
)
gx.test()
assignments = gx.assignments

Outputs

The results of geomux will be an assignment dataframe that has as many observations as there are input cells.

The columns of this dataframe will include:

Column Name	Description
cell_id	The name of the cell provided or the index.
assignment	A list representing all significant guides within that cell.
counts	The number of UMIs observed in the cell.
pvalues	The adjusted p-value of the hypergeometric test for that cell/guide test.
log_odds	The log odds of observing the highest scoring guide compared to the second highest.
moi	The number of significant guides within the cell.
n_umi	The number of UMIs observed in the cell.
min_pvalue	The minimum pvalue across all significant guides within the cell.
max_count	The maximum count across all significant guides within the cell.
tested	A bool flag representing whether the cell was included in the test (or False if it was filtered for low UMI counts)