scfair 0.6.0

Fair highly variable gene selection for single-cell RNA-seq

scFair

scFair selects highly variable genes (HVGs) for single-cell RNA-seq while giving small cell populations a fairer chance at the gene list.

Standard HVG ranking is driven by the largest populations. Markers that matter for rare types can fall just below the cutoff even when they are exactly what you need later for clustering or annotation. scFair keeps a familiar global ranking as the backbone and, by default, appends a small extension from the same ranking so near-miss genes are not discarded.

Docs: Read the Docs.

What you need

• Python 3.10+ (tested 3.10–3.12)
• An AnnData object with raw integer counts in .X or layers["counts"]

Install

pip install scfair

First run

import scfair as scf

# raw integer counts in .X or layers["counts"]
scf.pp.highly_variable_genes(adata)

adata = adata[:, adata.var["highly_variable"]].copy()

That is the default path: one global HVG pass, then a small append of the next genes in the same ranking. Runtime is on the same order as scanpy.pp.highly_variable_genes.

To let the data suggest how many genes to keep:

scf.pp.highly_variable_genes(adata, n_top_genes="auto")

To match scanpy with no extension:

scf.pp.highly_variable_genes(adata, balance_method="none")

Status

0.6.0 (Beta). Import as import scfair as scf and use names in scfair.__all__ and scf.pp. See the API reference.

Next steps

1. Installation
2. Quickstart
3. Tutorial: PBMC 10k misclustered cells vs standard HVG
4. FAQ if something looks off

License

Software: Apache License 2.0.

Author

Zhao Li (李钊)
Email: [e-mail]