[!WARNING] This is a major release with breaking changes. Samplesheets and command lines written for v2.0.0 will not work unmodified. See the migration notes below.

Added

• New samplesheet-driven input system: each row declares its own entry point via start_from, allowing subreads, CCS, full-length and long-read FASTA inputs to be mixed in a single run #40, #57
• Support for multiple libraries (cells) per sample, merged with TAMA
• Input chunking before mapping via the new chunker subworkflow, controlled by chunk_ccs and chunk_mapping
• Added parameter tama_merge_all to allow merging of sample-wise beds into a single annotation bed
• Updated nf-core template to version 4.1.0

Changed

• Breaking: the samplesheet format has been redefined. Columns are now sample,seq_data,pbi,start_from, replacing sample,bam,pbi,reads. The seq_data column takes any supported input file, and start_from (ccs, lima, refine or mapping) declares where that row enters the pipeline. The None placeholder is replaced by none for absent pbi files.
• Breaking: the chunk parameter has been split into chunk_ccs and chunk_mapping, which independently control chunking of CCS generation and of mapping.
• Breaking: the minimum Nextflow version is now 25.10.4 (was 23.04.0).
• Breaking: output files are now prefixed with <sample>_<N>, N being the 0-based row number in the samplesheet, so that several rows can share a sample name. Rows starting from lima, refine or mapping are split before mapping and their files are named <sample>_<N>.chunk.<X>.*. TAMA merge outputs keep the plain sample name. See the output documentation.
• Software versions are now collected through Nextflow topic channels instead of a dedicated module.
• --primers is no longer unconditionally required. It is only consumed by LIMA and isoseq refine, so it is now required only when at least one samplesheet row uses start_from ccs, lima or refine. Runs where every row uses mapping can omit it #50

Removed

• Breaking: removed the --entrypoint parameter. Per-row start_from in the samplesheet replaces it, and is strictly more flexible: entry points can now differ between samples within one run.
• Breaking: removed --max_cpus, --max_memory and --max_time. These were replaced by the resourceLimits directive in the nf-core template; set limits in a custom config instead.
• Removed the --hook_url parameter, following the nf-core template.
• Removed the custom/dumpsoftwareversions module.

Fixed

• Updated all nf-core modules to their latest revisions (bamtools/convert, gnu/sort, gstama/collapse, gstama/merge, gstama/polyacleanup, gunzip, isoseq/refine, lima, minimap2/align, multiqc, pbccs, ultra/align, ultra/index)
• Replaced the utils_nfvalidation_plugin subworkflow with utils_nfschema_plugin, moving parameter validation to nf-schema 2.5.1
• Moved local subworkflows into their own directories to follow nf-core guidelines
• Improved samplesheet validation: start_from is now an enum, file-extension patterns are correctly anchored, and a .bam.pbi index is required when start_from is ccs #48
• Fixed CSS / CCS typo in the documentation #36
• The documented ccs options min_passes, min_snr, top_passes, min_length and max_length are now actually passed to PBCCS (only rq was before)
• Removed a vulnerable artifact pattern from the PR-comment workflow #61

Dependencies

Deprecated

• None

• New entrypoint option to skip isoseq pre-processing.
• Update the pipeline to nf-core 2.14.1.
• Update modules.
• nf-validation version pinned PR25
• Upgrade from isoseq3 to isoseq (version 4) Fix segmentation fault PR27
• Add alternative entrypoint PR10

Added

• A new entreypoint system has been implemented to allow the user where to start the analysis.
• The isoseq entrypoint runs the full pipeline.
• The map entrypoint runs the pipeline from the mapping step.
• This new entreypoint option make possible to use the isoseq pipeline for analysis PacBio data when subreads are not provided, or for users who want to benefit from the mapping + TAMA analysis for their Nanopore data.

Fixed

• Update modules to their nf-test version (bamtools/convert, custom/dumpsoftwareversions, gnu/sort, gstama/collapse/ gstama/merge, gstama/polyacleanup, gunzip, isoseq/refine, lima, minimap2/align, pbccs,ultra/align, ultra/index)

• Since isoseq3 switch to version 4, it has been rename isoseq

  Tool	Previous version	New version
  bamtools/convert	2.5.2	2.5.2
  isoseq	3.8.2	4.0.0
  lima	2.7.1	2.9.0
  minimap2/align	2.24	2.28
  gnu/sort	8.25	9.3
  multiqc	1.21	1.24.1

• Update the pipeline to nf-core 2.9.

• Add gnu/sort to sort annotation before uLTRA index

• Update citations

• Add background to pipeline png

• Update modules

  Tool	Previous version	New version
  isoseq3	3.8.1	3.8.2
  lima	2.6.0	2.7.1
  bamtools/convert	2.5.1	2.5.2
  gstama/merge	1.0.2	1.0.3
  uLTRA/index	0.0.4.2	0.1
  uLTRA/align	0.0.4.2	0.1
  samtools	1.17	1.17
  gnu/sort	––	8.25

Fixed

• Update minimap2 path test: Don’t set gtf option. It is not expected to be used with minimap2 is chosen.
• FIX: Don’t prepare gtf channel when minimap2 is chosen.

Fixed

• Fix pipeline image path
• params.input invalid type if pipeline is run with local file in samplesheet (was working with URL)

Added

• Fix issue #17. Thanks to Husen M. Umer.
• Zenodo DOI
• Update to template v2.7.2

Fixed

• Remove hard coded capped option for GSTAMA_FILELIST step. Now follow user choice. Thanks to Mazdak Salavati.

Dependencies

Update the pipeline to nf-core template v2.5.1. Update modules:

• samplesheet_check
• dumpsoftwareversion
• MultiQC

Fix aligner option documentation.

Pipeline performance improvement. When uLTRA path is selected, It reduce computation by indexing GTF once instead of chunk times.

nf-core/isoseq is a bioinformatics best-practice analysis pipeline for Isoseq gene annotation with uLTRA and TAMA. Starting from raw isoseq subreads, the pipeline:

• Generates the Circular Consensus Sequences (CSS)
• Clean and polish CCS to create Full Length Non Chimeric (FLNC) reads
• Maps FLNCs on the genome
• Define and clean gene models