Hybrid assembly
What this exercise tests
You are given paired-end short reads and a long-read file for each sample. Combine both data types, assess the result and report whether the assembly is complete, circularised or still fragmented.
Before starting
Confirm that all three files belong to the same public sample identifier. Check the requirements in installing bioinformatics tools; the recommended GHRU-assembly hybrid route requires Linux or WSL2.
1. Inspect short reads
pixi run bactscout qc /absolute/path/to/short_reads \
-o /absolute/path/to/short_read_qc \
-t 8
Review quality, coverage and taxonomic composition before combining the reads.
2. Inspect long reads
For ONT R10 data:
pixi run bactscout long qc /absolute/path/to/long_reads \
--platform ont_r10
Select the platform value that matches the supplied data. Review read count, yield, N50, quality, coverage and taxonomy.
3. Prepare the GHRU-assembly sample sheet
sample_id,short_reads1,short_reads2,long_reads,genome_size
sample_01,/data/sample_01_R1.fastq.gz,/data/sample_01_R2.fastq.gz,/data/sample_01_long.fastq.gz,5.5m
Use absolute paths and keep one row per sample.
4. Run the hybrid workflow
git clone https://github.com/ghruproject/GHRU-assembly.git
cd GHRU-assembly
nextflow run main.nf \
--samplesheet /absolute/path/to/samplesheet.csv \
--outdir /absolute/path/to/hybrid_results \
-resume
The hybrid route uses Unicycler and runs QC around the assembly. Record the pipeline version, Nextflow version, container profile and non-default parameters. An equivalent hybrid workflow is acceptable unless one is specified by the exercise.
5. Judge completeness carefully
- single contig: a structural description only;
- circularised: supported by the assembly graph or validated terminal overlap;
- complete: expected replicons appear resolved and QC evidence is coherent;
- fragmented: unresolved sequence remains across multiple contigs.
Inspect the assembly graph, read mapping and coverage. Check whether small plasmids or other replicons may have been lost. Do not call a contig circular only because it is long.
Fill the result sheet
Record the public identifier, input filenames, assembler and version, species classification, overall QC outcome, polishing/circularisation evidence and any unresolved contamination, repeats or missing replicons.
Common problems
Short- and long-read taxonomy disagree: Stop and recheck sample identity.
Suspicious duplication at contig ends: Inspect terminal overlap and the assembly graph.
Small replicons are absent: Review long-read coverage and unassembled reads.
Polishing reduces quality: Confirm the correct reads, model and draft were used.