fingerling

The output of salmon alevin is a binary sparse matrix. fingerling is a python module for reading that output into a pandas dataframe suitable for downstream analysis.

Code is derived from k3yavi/vpolo, updated to use maturin for Python/Rust integration.

Installation

Requirements:

fingerling requires Python 3.10 (or greater) as well as:

pandas
numpy
scipy

Additionally, since fingerling contains a submodule written in Rust, a working installation of Rust is required. See instructions for installing Rust here.

Install fingerling from this repository using:

pip install git+https://github.com/milescsmith/fingerling

If you wish to download the source and build it locally, you will also need maturin. Prior to installing the local files, you will need to run maturin build followed by pip install .target/wheels/fingerling-X.X.X-cp3XX-cp3XX-OS_VERSION_PLATFORM.whl, where X.X.X corresponds to the package version, cp3XX the python version, and OS_VERSION_PLATFORM your operating system and version. The actual name and location for the wheel should be given by maturin at the end of the build.

Basic Usage

Currently, fingerling is not setup to be used as a standalone executable, but instead used as part of an analysis pipeline in something like a Jupyter notebook. The read_quants_bin function returns a sample-by-feature dataframe with an index corresponding to the cell barcodes and column names corresponding to the feature names and is suitable for use in creating an AnnData object:

import scanpy as sc
from fingerling import read_quants_bin

cite_df = read_quants_bin("/PATH/TO/ALEVIN_OUTPUT")

adata = sc.Anndata(X=cite_df)