hadge: a comprehensive pipeline for donor deconvolution in single-cell studies.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell multiplexing techniques like cell hashing and genetic multiplexing are crucial for optimizing sample processing and reducing costs in large-scale single-cell studies.
• Cell hashing involves conjugating antibody-tags or chemical-oligonucleotides to cell membranes for sample identification.
• Genetic multiplexing relies on aggregating RNA reads at known genomic coordinates to distinguish mixed samples.

Purpose of the Study:

• To develop an advanced computational pipeline for accurate deconvolution of multiplexed single-cell data.
• To integrate both cell hashing and genetic multiplexing deconvolution strategies for improved cell recovery.
• To validate the pipeline's performance on real-world biological samples, specifically fresh-frozen brain tissue.

Main Methods:

• Development of 'hadge' (hashing deconvolution combined with genotype information), a Nextflow pipeline.
• Integration of 12 distinct deconvolution algorithms for comprehensive analysis.
• Implementation of a joint deconvolution strategy combining the strengths of multiple methods.
• Application to nuclei hashing data from fresh-frozen brain tissue.

Main Results:

• The hadge pipeline successfully performs both hashing- and genotype-based deconvolution.
• A joint deconvolution strategy was proposed and implemented.
• This integrated approach led to the recovery of cells that were previously discarded by individual methods.
• Demonstrated improved cell recovery in nuclei hashing of fresh-frozen brain tissue.

Conclusions:

• The hadge pipeline provides a robust and integrated solution for single-cell multiplexing deconvolution.
• Joint deconvolution strategies significantly enhance cell recovery rates, improving data yield and quality.
• This approach is particularly valuable for analyzing challenging sample types like fresh-frozen tissues.