nf-core/pixelator

Edit

Introduction

This document describes the output produced by the pipeline. The directories listed below will be created in the results directory after the pipeline has finished. All paths are relative to the top-level results directory.

Pipeline overview

The pipeline is built using Nextflow and processes data using multiple subcommands of pixelator.

The pipeline consists of the following steps:

• Preprocessing
• Quality control
• Demultiplexing
• Duplicate removal and error correction
• Compute connected components
• Filtering, annotation, cell-calling
• Downstream analysis
• Generate reports

Preprocessing

The preprocessing step uses pixelator single-cell amplicon to create full-length amplicon sequences from both single-end and paired-end data. It returns a single fastq file per sample containing fixed length amplicons. This step will also calculate Q30 quality scores for different regions of the library.

Quality control

Quality control is performed using pixelator single-cell preqc and pixelator single-cell adapterqc.

The preqc stage performs QC and quality filtering of the raw sequencing data. It also generates a QC report in HTML and JSON formats. It saves processed reads as well as reads that were discarded (i.e. were too short, had too many Ns, or too low quality, etc.). Internally preqc uses Fastp, and adapterqc uses Cutadapt.

The adapterqc stage checks for the presence and correctness of the pixel binding sequences. It also generates a QC report in JSON format. It saves processed reads as well as discarded reads (i.e. reads that did not have a match for both pixel binding sequences).

Demultiplexing

The pixelator single-cell demux command assigns a marker (barcode) to each read. It also generates QC report in JSON format. It saves processed reads (one per antibody) as well as discarded reads with no match to the given barcodes/antibodies.

Duplicate removal and error correction

This step uses the pixelator single-cell collapse command.

The collapse command removes duplicate reads and performs error correction. This is achieved using the unique pixel identifier and unique molecular identifier sequences to check for uniqueness, collapse and compute a read count. The command generates a QC report in JSON format. Errors are allowed when collapsing reads if --algorithm is set to adjacency (this is the default option).

The output format of this command is an edge list in CSV format.

Compute connected components

This step uses the pixelator single-cell graph command. The input is the edge list dataframe (CSV) generated in the collapse step and after filtering it by count (--graph_min_count), the connected components of the graph (graphs) are computed and added to the edge list in a column called “component”.

The graph command has the option to recover components (technical multiplets) into smaller components using community detection to find and remove problematic edges. (See --multiplet_recovery). The information to keep track of the original and newly recovered components are stored in a file (components_recovered.csv).

Cell-calling, filtering, and annotation

This step uses the pixelator single-cell annotate command.

The annotate command takes as input the edge list (CSV) file generated in the graph command. It parses, and filters the edgelist to find putative cells, and it will generate a pxl file containing the edgelist, and an (AnnData object)[https://anndata.readthedocs.io/en/latest/] as well as some useful metadata.

Downstream analysis

This step uses the pixelator single-cell analysis command. Downstream analysis is performed on the pxl file generated by the previous stage. The results of the analysis is added to the pxl file.

Currently, the following analysis are performed:

• polarization scores (enable with --compute_polarization)
• co-localization scores (enable with --compute_colocalization)

Each analysis can be disabled by using respectively --compute_polarization false or --compute_colocalization false. This entire step can also be skipped using the --skip_analysis option.

Generate reports

This step uses the pixelator single-cell report command. This step will collect metrics and outputs generated by previous stages and generate a report in HTML format for each sample.

This step can be skipped using the --skip_report option.

More information on the report can be found in the pixelator documentation

Pipeline information

Nextflow provides excellent functionality for generating various reports relevant to the running and execution of the pipeline. This will allow you to troubleshoot errors with the running of the pipeline, and also provide you with other information such as launch commands, run times and resource usage.