downsample reads¶

At a glance

Repository: atlasxomics/downsample · Display name: downsample reads · Modality: Platform · Stage: Utility

Overview¶

A quick, platform-agnostic Workflow for downsampling paired-end reads to an exact target read count. Useful for producing lighter test sets or normalizing sequencing depth across samples before preprocessing. Input may be FASTA or FASTQ.

Steps¶

downsample_task — Runs bbmap/reformat.sh on the read pair with samplereadstarget=<out_reads>, writing downsampled files prefixed with ds_. Outputs are uploaded to latch:///downsampled/<out_dir>/.

Inputs¶

Parameter	Type	Description
`r1`	LatchFile	Read 1 (FASTA or FASTQ).
`r2`	LatchFile	Read 2 (FASTA or FASTQ).
`out_dir`	str	Subdirectory name under `downsampled/`. Cannot start with `/`.
`out_reads`	int	Exact number of output reads (or pairs) desired.

Outputs¶

A tuple of two LatchFiles — the downsampled ds_<r1> and ds_<r2> — written to latch:///downsampled/<out_dir>/.

downsampled/<out_dir>/
├── ds_<r1>
└── ds_<r2>

Example run¶

LaunchPlan(
    downsample,
    "Test Data",
    {
        "r1": LatchFile("latch:///.../D01033_NG01681_S3_L001_R1_001.fastq.gz"),
        "r2": LatchFile("latch:///.../D01033_NG01681_S3_L001_R2_001.fastq.gz"),
        "out_dir": "D01033_NG01681",
        "out_reads": 1000000,
    },
)