Requirements

Supported Python versions

The package consists predominantly of code written in Python, and the following versions are formally supported via continuous integration testing:

Python 3.8
Python 3.9
Python 3.10

Other versions may also work but have not been formally tested.

Software dependencies

The following auto_process subcommands depend on additional third-party software packages which must be installed separately:

Pipeline stage	Software packages	Notes
make_fastqs (+)	bcl2fastq 2.17	2.17+ recommended
make_fastqs	bcl-convert	Alternative to `bcl2fastq`
make_fastqs	cellranger	10xGenomics Chromium single-cell RNA-seq data only
make_fastqs	cellranger-atac	10xGenomics Chromium single-cell ATAC-seq data only
make_fastqs	cellranger-arc	10xGenomics Multiome ATAC + GEX data
make_fastqs	spaceranger	10xGenomics Visium spatial RNA-seq data only
run_qc (*)	fastqc
run_qc (*)	fastq_screen
run_qc (*)	bowtie	Required by fastq_screen
run_qc (*)	STAR	Required for strandedness and alignment
run_qc (*)	picard	Required for insert size metrics
run_qc (*)	rseqc	Required for gene body coverage
run_qc (*)	qualimap	Required for per-Fastq genomic origin of reads etc
run_qc	cellranger	10xGenomics Chromium single-cell RNA-seq data only
run_qc	cellranger-atac	10xGenomics single-cell ATAC-seq data only
run_qc	cellranger-arc	10xGenomics Multiome ATAC + GEX data
run_qc (*)	multiqc
run_qc (*)	seqtk	Required for protocols mapping subsequences of reads (e.g. `10x_Flex`)

(*) indicates packages that only need to be installed if Using conda to resolve pipeline dependencies hasn’t been enabled in the configuration (or by an appropriate command line option); otherwise the programs provided by these packages must be available on the PATH when the appropriate autoprocessor commands are issued. Using environment modules can be used to help manage this. Alternatively many of these packages can be obtained from the bioconda project.

(+) only needs to be installed if Using conda to resolve pipeline dependencies and the use_conda_for_bcl2fastq option (in the bcl_conversion section) haven’t been enabled in the configuration.

Note

Fastq generation requires Illumina’s bcl2fastq software. The recommended version is 2.17+ (earlier versions should work but note that they cannot handle NextSeq data); if there are multiple bcl2fastq packages available on the path at run time then see required_bcl2fastq_versions for how to specify which version is used.

Reference data

The following auto_process stages require additional reference data:

run_qc

run_qc

Reference data required for the calculation of various QC metrics are described in the sections below, along with the configuration settings needed to make them available to the QC pipeline.

FastqScreen

FastqScreen requires one or more conf files (each of which defines a specific “screen”) along with the underlying bowtie indexes for each organism which are included in the screens.

Indexes can be created manually, or by using the build_index.py utility (see Building indexes for aligners); the conf files must be created manually (see the FastqScreen documentation); each screen can then be added to the configuration using a screen section which references the corresponding conf file, e.g.:

[screen:model_organisms]
conf_file = /data/model_organisms.conf

The screens to use in the pipeline must be set using the fastq_screens parameter in the qc section, e.g.:

[qc]
fastq_screens = model_organisms,other_organisms,rRNA
...

Note

This replaces the old qc.setup script that was used to define the location of a set of standard screen conf files, used in earlier versions of the pipeline. Note that qc.setup is not longer needed (and will be ignored if present).

Strandedness

Strandedness determination requires STAR indexes for each organism of interest. These can be defined using appropriate settings in [organism:...] sections of the auto_process.ini file, for example:

[organism: human]
star_index = /data/genomeIndexes/hg38/STAR/

[organism: mouse]
star_index = /data/genomeIndexes/mm10/STAR/

Indexes can be created manually, or by using the build_index.py utility (see Building indexes for aligners).

Note

The [organism:...] sections supersede the old fastq_strand_indexes section of the auto_process.ini file; the older section is still recognised for now but is deprecated and likely to be dropped in future.

Insert size metrics (Picard)

Picard’s CollectInsertSizeMetrics needs a STAR index for each organism of interest (in order to generate a BAM file from the sequences). This should be specfied in the [organism:...] sections of the auto_process.ini configuration file, for example:

[organism: human]
star_index = /data/genomeIndexes/hg38/STAR/

STAR indexes can be created manually, or by using the build_index.py utility (see Building indexes for aligners).

RSeQC gene body coverage

RSeQC geneBody_coverage.py needs both a STAR index (in order to generate a BAM file from the sequences) and gene annotation in BED format, for each organism of interest. These should be specfied in the [organism:...] sections of the auto_process.ini configuration file, for example:

[organism: human]
star_index = /data/genomeIndexes/hg38/STAR/
annotation_bed = /data/genomeIndexes/hg38/hg38.HouseKeepingGenes.bed

Note

STAR indexes can be created manually, or by using the build_index.py utility (see Building indexes for aligners). Suitable gene model files for human and mouse can be downloaded from the RSeQC webpages at http://rseqc.sourceforge.net/#download-gene-models-update-on-12-14-2021

Qualimap RNA-seq metrics

Qualimap’s rnaseq command a STAR index (in order to generate a BAM file from the sequences) and gene annotation in GTF format, for each organism of interest. The pipeline uses RSeQC’s infer_experiment.py command to determine strand specificity for input to Qualimap.

All these should be specfied in the [organism:...] sections of the auto_process.ini configuration file, for example:

[organism: human]
star_index = /data/genomeIndexes/hg38/STAR/
annotation_gtf = /data/genomeIndexes/hg38/gencode.v40.annotation.gtf

STAR indexes can be created manually, or by using the build_index.py utility (see Building indexes for aligners).

Single cell analyses

Single library analyses of 10xGenomics single cell data require the appropriate compatible reference datasets for cellranger[-atac|-arc] count:

scRNA-seq data: transcriptome reference data set
snRNA-seq data: “pre-mRNA” reference data set (which includes both intronic and exonic information)
sc/snATAC-seq: Cell Ranger ATAC compatible genome reference
single cell multiome GEX+ATAC data: cellranger-arc compatible reference package

These can all be defined using appropriate settings in [organism:...] sections of the auto_process.ini file, for example:

[organism: human]
cellranger_reference = /data/10x/refdata-cellranger-GRCh38-1.2.0
cellranger_premrna_reference = /data/10x/refdata-cellranger-GRCh38-1.2.0_premrna
cellranger_atac_reference = /data/10x/refdata-cellranger-atac-GRCh38-1.0.1
cellranger_arc_reference = /data/10x/refdata-cellranger-arc-GRCh38-2020-A

[organism: mouse]
cellranger_reference = /data/10x/refdata-cellranger-mm10-1.2.0
cellranger_atac_reference = /data/10x/refdata-cellranger-atac-mm10-1.0.1
cellranger_arc_reference = /data/10x/refdata-cellranger-arc-mm10-2020-A

Note

Alternatively reference data sets can be specified at run-time for single cell and single nuclei RNA-seq using the --10x_transcriptome and --10x_premrna_reference command line options respectively with the run_qc command and the run_qc.py utility.

10xGenomics provide a number of reference data sets for scRNA-seq, ATAC-seq and single cell multiome data, which can be downloaded via:

There are also instructions for constructing reference data for novel organisms that are not supported by 10xGenomics.

Pre-mRNA references are currently not available, but the documentation explains how to generate a custom reference package for these data:

https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/advanced/references#premrna

Note

The [organism:...] sections supersede the old 10xgenomics... sections of the auto_process.ini file; the old sections are still recognised for now but are deprecated and likely to be dropped in future.

10x Genomics fixed RNA profiling (Flex) analyses

Analysis of 10xGenomics single cell fixed RNA profiling data (“Flex”) uses cellranger multi and requires:

Reference transcriptome dataset, and
Probe set data

These can be defined for specific organisms using the cellranger_reference and cellranger_probe_set settings in [organism:...] sections of the auto_process.ini file, for example:

[organism: human]
cellranger_reference = /data/10x/refdata-cellranger-GRCh38-1.2.0
cellranger_probe_set = /data/10x/Chromium_Mouse_Transcriptome_Probe_Set_v1.0_mm10-2020-A.csv

scRNA-seq data: transcriptome reference data set

Annotation data

Annotation data in BED and GTF formats can be specified for organisms of interest via the annotation_bed and annotation_gtf settings respectively in [organism:...] sections of the auto_process.ini file.

For example:

[organism: human]
annotation_bed = /data/genomeIndexes/hg38/annotation/hg38_NCBI_RefSeq_All.bed
annotation_gtf = /data/genomeIndexes/hg38/annotation/hg38_NCBI_RefSeq_All.gtf

[organism: mouse]
annotation_bed = /data/genomeIndexes/mm10/annotation/gencode.vM25.annotation.bed
annotation_gtf = /data/genomeIndexes/mm10/annotation/gencode.vM25.annotation.gtf

Building indexes for aligners

The _utilities_build_index.py utility can be used to build indexes for bowtie, bowtie2 and STAR from the appropriate data files (which must be obtained separately).

For example: to build indexes for hg38 using STAR version 2.7.7a:

build_index.py star -V 2.7.7a \
    -o hg38_STAR_2.7.7a_gencode40 \
    /mnt/genome_data/hg38/hg38.fa \
    /mnt/genome_data/hg38/hg38.gencode.v40.annotation.gtf

Note

If Using conda to resolve pipeline dependencies isn’t enabled then the required aligner must be accessible on the PATH, and the requested aligner version will be ignored.