10x Genomics
Chromium Single Cell Gene Expression

Cell Ranger3.1, printed on 11/21/2024

Specifying Input FASTQ Files for 10x Pipelines

The cellranger pipeline requires FASTQ files as input, which typically come from running cellranger mkfastq, a 10x-aware convenience wrapper for bcl2fastq. However, it is possible to use FASTQ files from other sources, such as Illumina's bcl2fastq, a published dataset, or our bamtofastq.

For experiments where only gene expression data is present, here are the arguments available for specifying which FASTQ files cellranger should use:

Argument	Brief Description
--fastqs	(Required) The folder containing the FASTQ files to be analyzed. Generally, this will be the fastq_path folder generated by cellranger mkfastq. If the files are in multiple folders, for instance because one library was sequenced across multiple flow cells, supply a comma-separated list of paths.
--sample	(Optional) Sample name to analyze. This will be as specified in the sample sheet supplied to mkfastq or bcl2fastq. Multiple names may be supplied as a comma-separated list, in which case they will be treated as one sample.
--libraries	Required for feature-barcoding analysis. Path to a libraries.csv file declaring input libraries. See Feature Barcoding Analysis page for details. Cannot be combined with --fastqs or --sample.
--lanes	(Optional) Lanes associated with this sample. Defaults to all lanes.
--indices	(Deprecated/Optional. Only used for output from cellranger demux.) Sample indices associated with this sample

For Feature Barcoding experiments, separate libraries for the gene expression reads and the Feature Barcoding reads are generated. In this case you must construct a csv file indicating the input data folder, sample name and library type of each input library, then pass this file to cellranger count using the --libraries flag. Please see the Feature Barcoding Analysis page for details on how to construct the libraries.csv file.

There is a wide range of ways bcl2fastq and mkfastq can be invoked, resulting in a wide range of potential file names and locations as the output. Since finding the right FASTQ files to process and the right arguments to process those files as desired can be confusing, we will illustrate some common scenarios.

Input FASTQ files should conform to the naming conventions of bcl2fastq and mkfastq, and are specified by providing the path to the folder containing them (via the --fastqs argument) and then optionally restricting the selection by specifying the samples and or lanes of interest.

To assist users, this page illustrates examples of how to handle common scenarios involving different FASTQ file folder hierarchies or naming conventions.

Quick Start:

Where are your FASTQ files?

• In an output folder from mkfastq or bcl2fastq (fastq_path) and:
  • In a subdirectory next to Reports and Stats folders, with expected sample name prefixes.
  • Within multiple folders per sample index, like "SI-GA-A1_1" and "SI-GA-A1_2".
  • In the same directory as the Reports and Stats folders.
• In a different folder:
  • I don't see Reports or Stats anywhere. The files are named like "MySample_S1_L001_I1_001.fastq.gz"

How are they named?

• Consistent with bcl2fastq/mkfastq, e.g. "mysample_S1_L001_I1_001.fastq.gz" (see above).
• Like "read-I1-AAAAAAA_lane-001-chunk-001.fastq.gz".
• Unlike any of the above examples.

Scenario: My FASTQs are in an output folder from mkfastq or bcl2fastq, in a subdirectory next to Reports and Stats folders, with expected sample name prefixes

How did I get here?

By running mkfastq with a simple CSV layout file or Illumina Experiment Manager samplesheet, or by running bcl2fastq directly (with an IEM samplesheet) on a flowcell. If you ran mkfastq, your files will be in a (MKFASTQ_ID)/outs/fastq_path folder, and your file hierarchy probably looks something like this:

MKFASTQ_ID
|-- MAKE_FASTQS_CS
`-- outs
    |-- fastq_path
        |-- HFLC5BBXX
            |-- test_sample1
            |   |-- test_sample1_S1_L001_I1_001.fastq.gz
            |   |-- test_sample1_S1_L001_R1_001.fastq.gz
            |   |-- test_sample1_S1_L001_R2_001.fastq.gz
            |   |-- test_sample1_S1_L002_I1_001.fastq.gz
            |   |-- test_sample1_S1_L002_R1_001.fastq.gz
            |   |-- test_sample1_S1_L002_R2_001.fastq.gz
            |   |-- test_sample1_S1_L003_I1_001.fastq.gz
            |   |-- test_sample1_S1_L003_R1_001.fastq.gz
            |   `-- test_sample1_S1_L003_R2_001.fastq.gz
            |-- test_sample2
            |   |-- test_sample2_S2_L001_I1_001.fastq.gz
            |   |-- test_sample2_S2_L001_R1_001.fastq.gz
            |   |-- test_sample2_S2_L001_R2_001.fastq.gz
            |   |-- test_sample2_S2_L002_I1_001.fastq.gz
            |   |-- test_sample2_S2_L002_R1_001.fastq.gz
            |   |-- test_sample2_S2_L002_R2_001.fastq.gz
            |   |-- test_sample2_S2_L003_I1_001.fastq.gz
            |   |-- test_sample2_S2_L003_R1_001.fastq.gz
            |   `-- test_sample2_S2_L003_R2_001.fastq.gz
        |-- Reports
        |-- Stats
        |-- Undetermined_S0_L001_I1_001.fastq.gz
        ...
        `-- Undetermined_S0_L003_R2_001.fastq.gz

If you ran bcl2fastq directly, then the output root folder would be where fastq_path is in the hierarchy above.

"Expected sample name prefixes" means you have one set of fastq files per sample, prefixed with the name of the sample as it appears in the simple CSV layout file or IEM samplesheet. Other situations described later on this page deal with the presence of four separate sets of files (four "samples" from bcl2fastq's point of view) per single biological sample/library.

For more information on the naming conventions, please visit Illumina's support site or refer to the bcl2fastq User Guide. The scenario where your files do not conform to the naming convention is described in a different section later on this page.

The table below describes the arguments you would pass into any analysis pipeline to target the right fastq files in this scenario. Be sure to substitute the capitalized text as appropriate. Also note that in most cases you will be passing a single sample into any given pipeline. Exceptions to this are described in the documentation for the individual pipelines. The "All Samples" entries in this table are provided for technical completeness.

Situation	Argument+Value
All samples (mkfastq)	--fastqs=MKFASTQ_ID/outs/fastq_path
All samples (mkfastq), multiple flowcells	--fastqs=MKFASTQ_ID/outs/fastq_path1,MKFASTQ_ID/outs/fastq_path2
All samples (bcl2fastq direct)	--fastqs=/PATH/TO/bcl2fastq_output
Process test_sample1 from all lanes (mkfastq)	--fastqs=MKFASTQ_ID/outs/fastq_path \ --sample=test_sample1
Process test_sample1 from lane 1 only (mkfastq)	--fastqs=MKFASTQ_ID/outs/fastq_path \ --sample=test_sample1 \ --lanes=1
Process test_sample1 and test_sample2 as a single merged sample (mkfastq)	--fastqs=MKFASTQ_ID/outs/fastq_path \ --sample=test_sample1,test_sample2

 

Scenario: My FASTQs are in an output folder from mkfastq or bcl2fastq, but there are multiple folders per sample index, like "SI-GA-A1_1" and "SI-GA-A1_2"

How did I get here?

It is likely that an input samplesheet was used that explicitly separated the four oligos in a 10x sample index set into four separate sample names. You may see a file hierarchy like this:

bcl2fastq_output
|-- HFLC5BBXX
    |-- SI-GA-A1_1
    |   |-- SI-GA-A1_1_S1_L001_I1_001.fastq.gz
    |   |-- SI-GA-A1_1_S1_L001_R1_001.fastq.gz
    |   `-- SI-GA-A1_1_S1_L001_R2_001.fastq.gz
    |-- SI-GA-A1_2
    |   |-- SI-GA-A1_2_S2_L001_I1_001.fastq.gz
    |   |-- SI-GA-A1_2_S2_L001_R1_001.fastq.gz
    |   `-- SI-GA-A1_2_S2_L001_R2_001.fastq.gz
    |-- SI-GA-A1_3
    |   |-- SI-GA-A1_3_S3_L001_I1_001.fastq.gz
    |   |-- SI-GA-A1_3_S3_L001_R1_001.fastq.gz
    |   `-- SI-GA-A1_3_S3_L001_R2_001.fastq.gz
    |-- SI-GA-A1_4
    |   |-- SI-GA-A1_4_S4_L001_I1_001.fastq.gz
    |   |-- SI-GA-A1_4_S4_L001_R1_001.fastq.gz
    |   `-- SI-GA-A1_4_S4_L001_R2_001.fastq.gz
|-- Reports
|-- Stats
|-- Undetermined_S0_L001_I1_001.fastq.gz
|-- Undetermined_S0_L001_R1_001.fastq.gz
`-- Undetermined_S0_L001_R2_001.fastq.gz

You probably want to be able to merge All samples from the SI-GA-A1 index into a single analysis. If you only run one index at a time, you will see a smaller number of reads than expected, which may translate to lower coverage or cell count than you expect for your experiment.

Situation	Argument+Value
All samples	--fastqs=MKFASTQ_ID/outs/fastq_path
Process all SI-GA-A1 reads in a single analysis	--fastqs=MKFASTQ_ID/outs/fastq_path \ --sample=SI-GA-A1_1,SI-GA-A1_2,SI-GA-A1_3,SI-GA-A1_4
Only process first sample index	--fastqs=MKFASTQ_ID/outs/fastq_path \ --sample=SI-GA-A1_1

 

Scenario: My FASTQs are in an output folder from mkfastq or bcl2fastq, in the same directory as the Reports and Stats folders

How did I get here?

An Illumina Experiment Manager-formatted samplesheet was used with either no entry or a blank entry for the Sample_Project column. Your hierarchy likely looks something like this:

fastq_path
|-- Reports
|-- Stats
|-- test_sample_S1_L001_I1_001.fastq.gz
|-- test_sample_S1_L001_R1_001.fastq.gz
|-- test_sample_S1_L001_R2_001.fastq.gz
|-- test_sample_S1_L002_I1_001.fastq.gz
|-- test_sample_S1_L002_R1_001.fastq.gz
|-- test_sample_S1_L002_R2_001.fastq.gz
|-- test_sample_S1_L003_I1_001.fastq.gz
|-- test_sample_S1_L003_R1_001.fastq.gz
|-- test_sample_S1_L003_R2_001.fastq.gz
|-- Undetermined_S0_L001_I1_001.fastq.gz
...
`-- Undetermined_S0_L003_R2_001.fastq.gz

This is fine; you would use the same arguments as if the FASTQs were organized into subfolders within the output folder.

Situation	Argument+Value
All samples (mkfastq)	--fastqs=MKFASTQ_ID/outs/fastq_path
All samples (bcl2fastq direct)	--fastqs=/PATH/TO/bcl2fastq_output
Process test_sample from all lanes (mkfastq)	--fastqs=MKFASTQ_ID/outs/fastq_path \ --sample=test_sample
Process test_sample from lane 1 only (mkfastq)	--fastqs=MKFASTQ_ID/outs/fastq_path \ --sample=test_sample \ --lanes=1

 

Scenario: My FASTQs are in a different folder; I don't see Reports or Stats anywhere. The files are named like "MySample_S1_L001_I1_001.fastq.gz"

How did I get here?

It is likely that FASTQ files have been transferred from either a mkfastq or bcl2fastq run into another folder. They still retain the names assigned by bcl2fastq, which is a combination of sample name, sample order, lane, read type, and chunk. Your file hierarchy may look like this:

PROJECT_FOLDER
|-- MySample_S1_L001_I1_001.fastq.gz
|-- MySample_S1_L001_R1_001.fastq.gz
|-- MySample_S1_L001_R2_001.fastq.gz
|-- MySample_S1_L002_I1_001.fastq.gz
|-- MySample_S1_L002_R1_001.fastq.gz
|-- MySample_S1_L002_R2_001.fastq.gz

This is fine; since the files are named according to the bcl2fastq standard, you would use the same arguments as if the FASTQs were organized into a flowcell folder or mkfastq output folder.

Situation	Argument+Value
All samples	--fastqs=/PATH/TO/PROJECT_FOLDER
Process MySample from all lanes	--fastqs=/PATH/TO/PROJECT_FOLDER \ --sample=MySample
Process MySample from lane 1 only	--fastqs=/PATH/TO/PROJECT_FOLDER \ --sample=MySample \ --lanes=1

 

My FASTQs are named like "read-I1-AAAAAAA_lane-001-chunk-001.fastq.gz"

How did I get here?

The 10x demux pipeline was used to demultiplex the flowcell instead of mkfastq. This pipeline has been deprecated, but you've still got a job to do. Your file hierarchy likely has many files in it, named as such:

demux_id
|-- BCL_PROCESSOR_CS
`-- outs
    |-- fastq_path
        |-- read-I1_si-AAAAAAAA_lane-001-chunk-001.fastq.gz
        ...
        |-- read-I1_si-TTTTTTTT_lane-002-chunk-001.fastq.gz
        |-- read-I1_si-X_lane-002-chunk-001.fastq.gz
        |-- read-RA_si-AAAAAAAA_lane-001-chunk-001.fastq.gz
        ...
        |-- read-RA_si-TTTTTTTT_lane-002-chunk-001.fastq.gz
        |-- read-RA_si-X_lane-002-chunk-001.fastq.gz

To ingest the correct FASTQ files from a demux run, you will need to know the 10x sample index or oligos associated with your sample. That will select the correct files from the sample indices in your folder:

Situation	Argument+Value
All samples	--fastqs=/PATH/TO/PROJECT_FOLDER
Process sample associated with SI-GA-A1	--fastqs=/PATH/TO/PROJECT_FOLDER \ --indices=SI-GA-A1
Process sample associated with SI-GA-A1, lane 1 only	--fastqs=/PATH/TO/PROJECT_FOLDER \ --indices=SI-GA-A1 \ --lanes=1
Process samples by sample index oligo	--fastqs=/PATH/TO/PROJECT_FOLDER \ --indices=AACCGTAA,CTAAACGG,GGTTTACT,TCGGCGTC

 

My FASTQs are not named like any of the above examples.

How did I get here?

It is likely that you received files that were processed through a proprietary LIMS system, which employs its own naming conventions.

10x pipelines need files named in the bcl2fastq or demux convention in order to run properly. You will need to determine which file corresponds to which sample and which read type, likely by consulting your sequencing core or the individual who demultiplexed your flowcell.

It is highly likely that these files were initially processed with bcl2fastq, so you will need to rename the files in the following format, once you track down their origin:

[Sample Name]_S1_L00[Lane Number]_[Read Type]_001.fastq.gz

Where Read Type is one of:

• I1: Sample index read (optional)
• R1: Read 1
• R2: Read 2

After you have renamed those files into that format, you'll use the following arguments:

Situation	Argument+Value
All samples	--fastqs=/PATH/TO/PROJECT_FOLDER
Process SAMPLENAME from all lanes	--fastqs=/PATH/TO/PROJECT_FOLDER \ --sample=SAMPLENAME
Process SAMPLENAME from lane 1 only	--sample=SAMPLENAME \ --fastqs=/PATH/TO/PROJECT_FOLDER \ --lanes=1