Universal preprocessing of single-cell genomics data

Abstract

We describe a workflow for preprocessing a wide variety of single-cell genomics data types. The approach is based on parsing of machine-readable seqspec assay specifications to customize inputs for kb-python, which uses kallisto and bustools to catalog reads, error correct barcodes, and count reads. The universal preprocessing method is implemented in the Python package cellatlas that is available for download at: https://github.com/cellatlas/cellatlas/.

Algorithms for a Commons Cell Atlas

“…To facilitate the generation of reference maps from single-cell genomics data, we developed a collection of tools, mx and ec , that operate on cell by feature (gene/isoform/protein/peak) matrices and marker equivalence class files respectively and work together with uniform preprocessing tools kallisto, bustools, kb-python , and cellatlas (Booeshaghi, Sullivan, and Pachter 2023; Melsted et al 2021; Melsted, Ntranos, and Pachter 2019; Bray et al 2016) (Methods). These tools solve key algorithmic and infrastructure problems in single-cell RNAseq preprocessing, namely automated cell type assignment, marker gene selection, and iterative data reprocessing.…”

“…First, data are often preprocessed with different tools introducing unnecessary computational variability (Davis et al 2018; Z. Zhang et al 2021; Booeshaghi, Sullivan, and Pachter 2023). Second, quantifications are often limited to the gene-level and do not distinguish between spliced and unspliced molecules (Hjörleifsson et al 2022).…”

kallisto, bustools, and kb-python for quantifying bulk, single-cell, and single-nucleus RNA-seq

“…one can specify -x 10xv3). ○ Option 2: One can use seqspec 40,41 which contains machine-readable specifications for a wide range of sequencing assays. ○ Option 3: One can format their own custom technology string specifying the read locations of the barcodes, unique molecular identifiers (UMIs), and the biological sequence that is to be mapped (Box 3).…”

“…one can specify -x 10xv3). ○ Option 2: One can use seqspec 40,41 which contains machine-readable specifications for a wide range of sequencing assays.…”

“…The technology string can be used in one of three ways: Option 1: Several assays are predefined within the software (the list is viewable by calling kb --list ) so one can name one of those directly (e.g. one can specify -x 10xv3). Option 2: One can use seqspec 40,41 which contains machine-readable specifications for a wide range of sequencing assays. Option 3: One can format their own custom technology string specifying the read locations of the barcodes, unique molecular identifiers (UMIs), and the biological sequence that is to be mapped ( Box 3 ). Strandedness : If a read (or the first read in the case of paired-end reads) is to be mapped in forward orientation, one should specify --strand=forward . If it is to be mapped in reverse orientation, one should specify --strand=reverse .…”