nf-core pipeline · isoseq, isoseq-3, rna
nf-core/isoseq
A Nextflow pipeline for Iso-Seq genome annotation, generating FLNC reads and gene models using uLTRA and TAMA.
What nf-core/isoseq is
nf-core/isoseq is a bioinformatics analysis pipeline designed for genome and transcript isoform annotation using Pacific Biosciences (PacBio) Iso-Seq long-read sequencing data. It was originally written by Sébastien Guizard and developed as part of the European Union Horizon 2020 GENE-SWitCH project. The workflow accepts long-read transcriptomic datasets starting from raw subreads or intermediate processing stages to define, polish, and annotate gene models.
Long-read Iso-Seq sequencing enables the observation of full-length transcripts without computational assembly, but processing the data requires complex, multi-stage pipelines to build consensus reads, strip artificial adapters, discard chimeric sequences, and accurately align spliced reads across genomes. nf-core/isoseq standardises these steps into an automated workflow, producing clean gene models and genome annotations in BED format using established tools like uLTRA and TAMA.
What the pipeline does
The pipeline processes reads through several discrete stages depending on the chosen starting point. For raw subreads, Circular Consensus Sequencing (CCS) consensus reads are generated with PBCCS, followed by primer removal using LIMA. Non-chimeric sequences are identified and isolated using ISOSEQ3 REFINE, BAM files are converted to FASTA format with BAMTOOLS CONVERT, and polyA-tail containing reads are filtered and trimmed via GSTAMA_POLYACLEANUP to yield polished Full Length Non-Chimeric (FLNC) sequences.
Next, FLNC reads are mapped to a reference genome using either MINIMAP2 or uLTRA (which indexes an input GTF file and decompresses the reads beforehand). Once aligned, transcript models are collapsed and filtered using TAMA collapse, and the final gene annotations are merged across samples using TAMA merge.
Key facts
- Pipeline engine
- Nextflow DSL2
- Latest release
- 2.0.0 (2024-09-05)
- Original developer
- Sébastien Guizard
- Input formats
- BAM (.bam), PacBio Index (.pbi), compressed FASTA (.fa.gz)
- Input read types
- Raw PacBio subreads, Circular Consensus Sequences (CCS), LIMA full-length sequences, PacBio HiFi, and Oxford Nanopore long reads
- Output format
- BED annotation
- Aligners supported
- uLTRA, MINIMAP2
- Transcript collapsing tool
- TAMA
What you can use it for
- De novo gene and isoform annotation
- Annotate full-length transcript isoforms and novel gene models from raw PacBio Iso-Seq subreads mapped against a reference genome.
- Processing pre-computed consensus reads
- Run downstream primer trimming, chimera refinement, and genome mapping on existing CCS BAM files without recomputing consensuses.
- Long-read alignment and model merging
- Align pre-filtered PacBio HiFi or Oxford Nanopore long reads directly using MINIMAP2 or uLTRA, collapsing redundant models and merging multi-sample annotations via TAMA.
Run it here
The research agent runs nf-core/isoseq with Nextflow on its own machine — up to 32 cores, 256 GiB of memory and 24 hours per run, which covers the test profile and real single-sample runs. It builds the samplesheet, checks it against the pipeline's schema and reads the results back. When a run is too big for one machine, it gives you the exact command for your own cluster instead.
Strengths and limits
Good at
- Accepts inputs at multiple entry points along the workflow, ranging from raw subreads to pre-processed consensus or mapping-ready FASTA files.
- Accommodates both PacBio Iso-Seq data and Oxford Nanopore long reads for downstream alignment and annotation.
- Integrates both MINIMAP2 and splice-aware uLTRA aligners alongside TAMA collapse and merge algorithms.
- Maintains individual software dependencies in isolated Docker and Singularity containers through Nextflow DSL2 modules.
Watch out for
- Starting from raw subread BAM files requires accompanying PacBio Index (.pbi) files.
- Utilising the uLTRA alignment path requires an existing reference GTF file alongside the genome sequence.
Questions
Can I run nf-core/isoseq without my own cluster?
The research agent runs nf-core/isoseq with Nextflow on its own machine — up to 32 cores, 256 GiB of memory and 24 hours per run, which covers the test profile and real single-sample runs. It builds the samplesheet, checks it against the pipeline's schema and reads the results back. When a run is too big for one machine, it gives you the exact command for your own cluster instead.
What licence is nf-core/isoseq under?
MIT, like every nf-core pipeline.
Can this pipeline be run on Oxford Nanopore data?
Yes. Oxford Nanopore long-read sequences can be supplied directly at the mapping entry point in compressed FASTA format.
What alignment tools are available in the pipeline?
Users can align full-length reads to the genome using either MINIMAP2 or uLTRA.
Do I have to start the analysis from raw PacBio subreads?
No. The pipeline supports flexible entry points via the sample sheet, allowing execution to begin at the CCS, LIMA, refine, or mapping steps.
What is the final annotation output?
The pipeline collapses and merges gene models across samples using TAMA, producing an annotated BED file.
How to cite it
Sébastien Guizard, Katarzyna Miedzinska, Jacqueline Smith, Jonathan Smith, Richard I Kuo, Megan Davey, Alan Archibald & Mick Watson. nf-core/isoseq: simple gene and isoform annotation with PacBio Iso-Seq long-read sequencing. Bioinformatics, Volume 39, Issue 5, May 2023. doi: 10.1093/bioinformatics/btad150
Written from nf-core/isoseq's own README. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.