target-asm

Target eukaryotic genome assembly from contaminated PacBio HiFi read sets.

target-asm is a Nextflow DSL2 workflow for recovering target eukaryotic genomes from mixed long-read sequencing libraries. It is intended for samples where the organism of interest is sequenced together with host, symbiont, microbial, environmental, or culture-associated DNA.

The workflow is reference-independent with respect to the target genome. It first assembles the full read set as a metagenome, removes non-target contigs with NCBI FCS-GX, recruits original reads that support the target-enriched draft, optionally downsamples those reads, reassembles with hifiasm, and runs a final FCS-GX screen.

The motivating benchmark was contaminated PacBio HiFi sequencing of obligate biotrophic oomycetes, but the strategy is not oomycete-specific. Any eukaryotic target with an appropriate NCBI taxonomy ID for FCS-GX can be used.

Workflow

1. metaMDBG assembles the input HiFi reads as a metagenome.
2. FCS-GX removes contigs outside the requested target taxon.
3. minimap2 maps the original HiFi reads back to the cleaned draft.
4. samtools extracts mapped reads for target-enriched reassembly.
5. rasusa optionally downsamples mapped reads to a target base count.
6. hifiasm reassembles the recruited reads.
7. FCS-GX runs a final contamination screen.
8. Optional Compleasm and QUAST reports track assembly quality across steps.

Requirements

• Nextflow
• One of Docker, Singularity, or Apptainer
• PacBio HiFi reads in fastq.gz format
• An NCBI FCS-GX database
• An NCBI taxonomy ID for the target clade

All workflow tools are configured as containers in nextflow.config.

Get the Workflow

git clone https://github.com/Yixuan39/target-asm.git
cd target-asm
nextflow run main.nf --help

Quick Start

nextflow run main.nf \
  --reads /path/to/sample.fastq.gz \
  --gx_db /path/to/fcs-gx-db-prefix \
  --tax_id <target_ncbi_tax_id> \
  --target_bases <expected_genome_size> \
  --outdir results \
  -profile apptainer

Set --tax_id to the NCBI taxonomy ID for the target organism or target clade. Adjust --target_bases to the expected genome size multiplied by the desired coverage. For example, a 90 Mb genome at 60x coverage is 5.4e9.

To use all mapped reads for hifiasm reassembly, omit --target_bases.

When to Use target-asm

Use target-asm when contamination is too complex for read-length filtering or whole-library assembly alone. It is designed for cases where:

• the target is a eukaryote represented by a minority or mixed fraction of reads;
• contaminant reads overlap the target reads in length or quality;
• a close reference genome is unavailable or should not drive assembly;
• taxonomic cleaning plus read recruitment is preferable to manual contig filtering.

For clean single-organism HiFi libraries, running hifiasm directly is usually enough.

Parameters

Parameter	Required	Default	Description
--reads	yes	null	PacBio HiFi reads, compressed as .fastq.gz.
--gx_db	yes	null	Path prefix for the NCBI FCS-GX database.
--tax_id	yes	null	NCBI taxonomy ID used as the target taxon for FCS-GX.
--target_bases	no	null	Number of bases to retain with rasusa before hifiasm reassembly.
--rasusa_seed	no	0	Random seed for rasusa.
--threads	no	24	CPU threads used by threaded processes.
--fcs_gx_memory	no	512 GB	Memory requested for each FCS-GX process.
--hifiasm_option	no	-l 2	Extra options passed to hifiasm.
--outdir	no	.	Output directory for published files.
--keep_intermediates	no	false	Publish intermediate assemblies and mapped reads.
--quality_library	no	null	Compleasm database path. Enables assembly QC when supplied.
--quality_lineage	with QC	''	Compleasm lineage name.
--help	no	false	Print command-line help.

Quality Control

Add Compleasm and QUAST summaries to a full workflow run:

nextflow run main.nf \
  --reads /path/to/sample.fastq.gz \
  --gx_db /path/to/fcs-gx-db-prefix \
  --tax_id <target_ncbi_tax_id> \
  --target_bases 5.4e9 \
  --quality_library /path/to/compleasm_db \
  --quality_lineage <busco_lineage> \
  --outdir results \
  -profile apptainer

When QC is enabled, target-asm evaluates the metaMDBG assembly, the first FCS-GX-cleaned assembly, the hifiasm assembly, and the final FCS-GX-cleaned assembly.

Outputs

With the default settings, the main output is:

results/
  <sample>.fasta.gz
  fcs_gx/
    fcs_initial.fcs_gx_report.txt
    fcs_final.fcs_gx_report.txt

When --quality_library and --quality_lineage are provided:

results/
  quality/
    quality_trace.csv
    quality_final.csv

When --keep_intermediates is set, target-asm also publishes intermediate outputs under metaMDBG/, minimap2/, rasusa/, hifiasm/, and fcs_gx/.

Profiles

The workflow includes these profiles:

Profile	Description
standard	Local execution.
slurm	SLURM execution.
docker	Enable Docker containers.
singularity	Enable Singularity containers with automounts.
apptainer	Enable Apptainer containers with automounts.

Profiles can be combined, for example -profile slurm,apptainer, when running on a SLURM cluster. On SLURM systems, the slurm profile is recommended because only the FCS-GX steps require high-memory nodes, while the remaining workflow steps can run with ordinary scheduler resources.

Notes

• NCBI recommends 512 GiB shared memory for FCS-GX with the standard database; running below this can be extremely slow. target-asm requests --fcs_gx_memory '512 GB' by default.
• target-asm removes non-target taxonomic contamination, but target-derived organellar contigs may remain and should be handled downstream if nuclear-only assemblies are required.
• For the downy mildew benchmark, Oomycota was used as the target clade (--tax_id 4762) and stramenopiles was used for Compleasm QC. For other targets, choose the matching NCBI taxon and Compleasm/BUSCO lineage.