10x Genomics
Chromium Single Cell Gene Expression

Cell Ranger7.1, printed on 11/21/2024

Creating a Reference Package with cellranger mkref

Table of contents

• Cell Ranger mkref
• Compatible use cases
• Making a reference package
• Filter with mkgtf
• Index with cellranger mkref
• Single species
• Multiple species
• Adding one or more genes to your reference
• Generating a Cell Ranger compatible "pre-mRNA" reference package
• Check for multi-mapped reads

Cell Ranger mkref

Cell Ranger provides pre-built human, mouse, and barnyard (human & mouse) reference packages for read alignment and gene expression quantification in cellranger count. Build notes are available here.

Cell Ranger allows users to create a custom reference package using cellranger mkref. To make a custom reference, you will need a reference genome sequence (FASTA file) and gene annotations (GTF file). A tutorial, Build a Custom Reference (cellranger mkref), is available to walk you through the steps.

Custom references built with previous versions of cellranger mkref can be used with the latest versions of cellranger count or cellranger multi. However, references built with the latest cellranger mkref may not be compatible with all older versions of the pipelines.

Compatible use cases

Cell Ranger supports the use of customer-generated references under the following conditions:

• Your reference should have only a small number of overlapping gene annotations. Reads aligning non-uniquely to multiple genes cause the pipeline to detect fewer molecules.
• Your FASTA and GTF files must be compatible with the open source splicing-aware RNA seq aligner, STAR. To be considered for transcriptome alignment, genes must have annotations with feature type 'exon' (column 3) in the GTF file.

Example use cases:

• Create reference for single species or multiple species
• Add a gene to an existing reference package
• Create custom reference for single-nuclei RNA-seq

Making a reference package

To create a custom reference:

1. Filter GTF file with mkgtf to contain only genes of interest
2. Index the FASTA and GTF files with mkref

Filter with mkgtf

GTF files downloaded from sites like ENSEMBL and UCSC often contain transcripts and genes which need to be filtered from your final annotation. Cell Ranger provides mkgtf, a simple utility to filter genes based on their key-value pairs in the GTF attribute column. The command syntax requires input and output GTF file names and --attribute values specifying gene biotypes to filter from the GTF file (replace values in red):

cellranger mkgtf input.gtf output.gtf --attribute=key:allowable_value

In the command above, the allowable_value can be any of the accepted biotypes listed below:

• protein_coding
• lncRNA
• antisense
• IG_C_gene
• IG_D_gene
• IG_J_gene
• IG_LV_gene
• IG_V_gene
• IG_V_pseudogene
• IG_J_pseudogene
• IG_C_pseudogene
• TR_C_gene
• TR_D_gene
• TR_J_gene
• TR_V_gene
• TR_V_pseudogene
• TR_J_pseudogene

For example, the following filtering was applied to generate the GTF file for the GRCh38 Cell Ranger reference package:

cellranger mkgtf Homo_sapiens.GRCh38.ensembl.gtf Homo_sapiens.GRCh38.ensembl.filtered.gtf \
                   --attribute=gene_biotype:protein_coding \
                   --attribute=gene_biotype:lncRNA \
                   --attribute=gene_biotype:antisense \
                   --attribute=gene_biotype:IG_LV_gene \
                   --attribute=gene_biotype:IG_V_gene \
                   --attribute=gene_biotype:IG_V_pseudogene \
                   --attribute=gene_biotype:IG_D_gene \
                   --attribute=gene_biotype:IG_J_gene \
                   --attribute=gene_biotype:IG_J_pseudogene \
                   --attribute=gene_biotype:IG_C_gene \
                   --attribute=gene_biotype:IG_C_pseudogene \
                   --attribute=gene_biotype:TR_V_gene \
                   --attribute=gene_biotype:TR_V_pseudogene \
                   --attribute=gene_biotype:TR_D_gene \
                   --attribute=gene_biotype:TR_J_gene \
                   --attribute=gene_biotype:TR_J_pseudogene \
                   --attribute=gene_biotype:TR_C_gene

This generated a filtered GTF file Homo_sapiens.GRCh38.ensembl.filtered.gtf from the original unfiltered GTF file Homo_sapiens.GRCh38.ensembl.gtf. In the output file, other biotypes such as gene_biotype:pseudogene are excluded from the GTF annotation.

Index with cellranger mkref

To create custom references, use the cellranger mkref command, passing it one or more matching sets of FASTA and GTF files. This utility copies your FASTA and GTF, indexes these in several formats, and outputs a folder with the name you pass to --genome. Input GTF files are typically filtered with mkgtf prior to mkref.

Argument	Description
--genome	Required. Unique genome name(s), used to name output folder. Should contain only alphanumeric characters and optionally period, hyphen, and underscore characters [a-zA-Z0-9_-]+. Specify multiple genomes by specifying the --genome argument multiple times.
--fasta	Required. Path(s) to FASTA file containing your genome reference. Specify multiple genomes by specifying the --fasta argument multiple times.
--genes	Required. Path(s) to GTF file(s) containing annotated genes for your genome reference. Specify multiple genomes with the --genes argument for each genome.
--memgb	Optional. Maximum memory (GB) used during STAR genome index generation. Defaults to 16. Please note, the amount of memory specified must be greater than the number of gigabases in the input reference FASTA file.
--ref-version	Optional. Reference version string to include with reference.
--nthreads	Optional. Number of threads used during STAR genome index generation. Defaults to 1.
--help or -h	Optional. Show list of all arguments and options.
--version	Optional. Show version.

Basic usage

cellranger mkref --genome=output_genome --fasta=input.fa --genes=input.gtf

System requirements

Indexing a typical human 3Gb FASTA file often takes up to 8 core hours and requires 32 GB of memory.

Outputs

A successful mkref run should conclude with a message similar to this:

Creating new reference folder at output_genome
...done
 
Writing genome FASTA file into reference folder...
...done
 
Computing hash of genome FASTA file...
...done
 
Writing genes GTF file into reference folder...
WARNING: The following transcripts appear on multiple chromosomes in the GTF:
 
This can indicate a problem with the reference or annotations. Only the first chromosome will be counted.
...done
 
Computing hash of genes GTF file...
...done
 
Writing genes index file into reference folder (may take over 10 minutes for a 3Gb genome)...
...done
 
Writing genome metadata JSON file into reference folder...
...done
 
Generating STAR genome index (may take over 8 core hours for a 3Gb genome)...
...done.
 
>>> Reference successfully created! <<<

Output listing:

── output_genome
    ├── fasta
    │   ├── genome.fa
    │   └── genome.fa.fai
    ├── genes
    │   └── genes.gtf.gz
    ├── reference.json
    └── star
        ├── chrLength.txt
        ├── chrNameLength.txt
        ├── chrName.txt
        ├── chrStart.txt
        ├── exonGeTrInfo.tab
        ├── exonInfo.tab
        ├── geneInfo.tab
        ├── Genome
        ├── genomeParameters.txt
        ├── SA
        ├── SAindex
        ├── sjdbInfo.txt
        ├── sjdbList.fromGTF.out.tab
        ├── sjdbList.out.tab
        └── transcriptInfo.tab

Single species

The most common use case is to create a reference for only one species. In this case, there is one set of matched FASTA and GTF files typically obtained from Ensembl, NCBI, or UCSC.

cellranger mkref --genome=GRCh38 --fasta=GRCh38.fa --genes=GRCh38-filtered-ensembl.gtf

When possible, please obtain genome sequence (FASTA) and gene annotations (GTF) from the same source: Use Ensembl FASTA files with Ensembl GTF files. Chromosome or sequence names in the FASTA file must match the chromosome or sequence names in the GTF file.

As noted in the STAR manual, the most comprehensive genome sequence and annotations are recommended:

• For the genome sequence, include all major chromosomes, unplaced and unlocalized scaffolds, but do not include patches and alternative haplotypes.
  • In Ensembl, the recommended genome file to download is annotated as "primary assembly."
  • In NCBI, it is "no alternative - analysis set."
• For the GTF file, genes must be annotated with feature type "exon" (column 3).
  • Prior to mkref, GTF annotation files from Ensembl and NCBI are typically filtered with mkgtf to include only a subset of the annotated gene biotypes.

Multiple species

To create a reference for multiple species, run the mkref command with multiple FASTA and GTF files. This is similar to the single species case above, but note that the order of the arguments matters. The arguments are grouped by the order they appear; for instance, the first --genome option listed corresponds to the first --fasta and --genes options listed. Please use or create this type of reference when analyzing barnyard validation experiments for estimating multiplet rates.

cellranger mkref --genome=GRCh38 --fasta=GRCh38.fa --genes=GRCh38-filtered-ensembl.gtf \
                 --genome=mm10 --fasta=mm10.fa --genes=mm10-filtered-ensembl.gtf

Adding one or more genes to your reference

Provided that you follow the format described above, it is fairly simple to add custom gene definitions to an existing reference. First, add the additional FASTA sequence records to the fasta/genome.fa file. Next, update the GTF file, genes/genes.gtf, with the gene annotation record(s). An example is described in the cellranger mkref tutorial for adding a marker gene to the FASTA and GTF files.

The GTF file format is essentially a list of records, one per line, each comprising nine tab-delimited non-empty fields.

After adding the necessary records to your FASTA file and the additional lines to your GTF file, run cellranger mkref as normal.

Generating a Cell Ranger compatible "pre-mRNA" reference package

The single-nuclei RNA-seq assay captures unspliced pre-mRNA as well as mature mRNA. However, after alignment, cellranger count only counts reads aligned to exons. Since the pre-mRNA will generate intronic reads, it may be useful to count these reads as well. Previously, it was recommended to create a custom “pre-mRNA” reference package, listing each gene transcript locus as an exon, in order to count intronic reads. In Cell Ranger 5.0, there is a new include-introns option for counting intronic reads that should be used instead, and the usage of pre-mRNA references is deprecated.

Check for multi-mapped reads

A read may align to multiple transcripts and genes, but Cell Ranger only considers a read confidently mapped to the transcriptome if it is mapped to a single gene (after converting the xf tag value to binary, 1-bit means the read is confidently mapped to the transcriptome).

To assess whether reads mapped to multiple genes, examine the GX or GN tags in the output BAM file, which are generated by Cell Ranger after alignment with STAR. Uniquely mapped reads will have one gene ID for GX and one gene name for GN , while multi-mapped reads will list multiple gene IDs and names.