nf-core/riboseq v2.0.0 - Nimble Narwhal

This is a major release. nf-core/riboseq grows from a single-caller ORF-detection pipeline into a full cross-caller ORF discovery and quantification suite, with novel-transcript discovery and ORF-level differential translation analysis layered on top of the existing gene-level pipeline.

Novel transcript discovery and extended ORF analysis

• Discover novel transcripts from your own RNA-seq data with StringTie, or supply your own via --novel_gtf. Novel transcripts are merged with a one-transcript-per-gene canonical backbone (--canonical_gtf) into a hybrid reference GTF.
• --extended_orf_analysis routes that hybrid GTF into the genome-BAM ORF callers (and a second STAR pass against the hybrid transcriptome for RiboCode), so novel intergenic ORFs are called alongside annotated ones.

More ORF callers

• Two new opt-in ORF callers: Rp-Bp (--run_rpbp) and PRICE (--run_price), both Bayesian and both flagged opt-in because they’re slow at genome scale.
• Ribotricer is demoted from a default caller to opt-in (--run_ribotricer, replacing --skip_ribotricer) after we found its per-ORF scores unstable across replicates; the default caller set is now Ribo-TISH + RiboCode.

Cross-caller ORF catalogue

• Under --extended_orf_analysis, every enabled caller’s predictions are normalised into a unified BED12 cohort catalogue, clustered by transcript anchor (annotated ORFs) or reciprocal exon overlap (novel intergenic ORFs), with optional smORF peptide collapse (--skip_orf_collapse) and a consensus view (--orf_min_callers / --orf_min_samples).
• Per-ORF in-frame P-site quantification produces an ORF x sample count matrix, and ORF-level differential translation (anota2seq / deltaTE / DOTSeq) runs on top of it, alongside the existing gene-level DTE.

Translational efficiency and quantification

• New --te_quantification_method plastid_psite (in-frame P-site counts) is now the default, replacing alignment-based (STAR + Salmon) counts. Pass --te_quantification_method alignment to restore the previous behaviour.
• --pseudo_aligner now actually selects the pseudo-alignment tool used by --te_quantification_method pseudo (previously always Salmon regardless of setting); kallisto is a working alternative alongside Salmon.
• New bigWig coverage tracks (strand-aware, IGV/JBrowse-ready), riboWaltz QC plots wired into MultiQC, and a --ribo_removal_tool choice of SortMeRNA (default), Bowtie2, or RiboDetector for rRNA removal.
• Per-sample UMI handling via an optional with_umi samplesheet column, and optional read-length equalisation to trim RNA-seq reads to match Ribo-seq lengths before quantification.

Breaking changes

• Output directories are reorganised into category-based folders for alignment, reference indexes, ORF prediction, quantification, transcript assembly and translational efficiency.
• Default --te_quantification_method changes from alignment to plastid_psite (see above) - per-gene counts are not comparable across the two.
• orf_class is now purely positional and smORF is gone (use the new is_smorf column instead); several orf_class values are renamed (see docs/output.md for the full vocabulary), and catalogue row counts shift as a result of the new clustering and collapse logic.
• --min_mapped_reads, --skip_pseudo_alignment and --skip_alignment are removed (none of them worked); --skip_ribotricer is replaced by --run_ribotricer.
• Minimum Nextflow version raised to 25.10.4.

Every parameter addition/removal and dependency bump is tracked in the full changelog, along with ~30 bug fixes accumulated since v1.2.0.

Credits

Special thanks to everyone who contributed to this release, including those who reviewed the release PR: @bediagan, @vagkaratzas, @FelixKrueger, @FriederikeHanssen, @iraiosub, @JackCurragh, @Joon-Klaps, @pinin4fjords, @luisas, @mashehu, @maxulysse, @suhrig.

Full Changelog: https://github.com/nf-core/riboseq/compare/1.2.0...2.0.0