v1.1.0 - Copper Hairpin

Contributors: @Vicbeg. Reviewers: @vagkaratzas, @piplus2, @nschan, @d4straub.

Behaviour change: treated groups whose sample_group shares no leading substring with any control now get no control; the reference-wide lone-control fallback is gone. (#16)

Added

• #19 - --four_base_dms (default true) applies four-base DMS analysis (Mitchell et al., 2023; RNA Framework 2.9.8) to DMS-MaP samples at pH ≥ 8: rf-count keeps only G→C/G→T mutations on G and drops indels, and rf-norm and rf-normfactor normalise each base separately over the whole transcript. These samples always use Mitchell normalisation (--rfnorm_norm_method is ignored), and --rfnorm_score_method 1 or 2 is rejected at launch. Set it to false for the previous all-mutation, single-scale behaviour.
• #19 - --rffold_use_alifold (default true) folds the replicates of a multi-replicate group together with RNAalifold (rf-fold -ua, new in RNA Framework 2.9.8), giving one consensus structure constrained by all their reactivity profiles; set it to false for the previous per-replicate RNAfold folding.
• #16 - --cutadapt_nextseq_trim passes Cutadapt --nextseq-trim with the given 3′ quality cutoff, stripping the high-quality poly-G tails that 2-colour (NextSeq/NovaSeq) chemistry leaves behind and quality trimming misses.
• #16 - --cutadapt_discard_untrimmed drops reads in which no adapter was found, for short-insert libraries (e.g. tRNA, miRNA) where every genuine read runs into the adapter. It cannot be combined with --cutadapt_quality_only.
• #16 - --rfcount_map_max_coverage passes rf-count's --max-coverage for MaP samples, downsampling each transcript above the cap (minimum 1000), so a transcript at extreme depth (e.g. a viral genome at ~10⁶×) counts in hours rather than days. Off by default. Ignored with --count_genome true.

Changed

• #19 - With RNA Framework 2.9.8, rf-fold draws its SVG structure plots itself, so the pipeline no longer passes it -vrp (--rffold_vienna_rnaplot still sets RNAplot for the ViennaRNA and rf-structextract plots).
• #19 - Results change for existing analyses because the two new options above default to true: DMS-MaP samples at pH ≥ 8 get four-base DMS reactivities, and groups with more than one replicate get one RNAalifold consensus structure instead of per-replicate RNAfold structures. Set --four_base_dms false and --rffold_use_alifold false to reproduce earlier releases.
• #16 - Fuzzy control pairing now picks the untreated or denatured control whose sample_group shares the longest leading substring with the treated group, rather than matching on the first substring only, so each arm of a design such as HFF_uninfected / HFF_HCMV_05hpi / HFF_HCMV_72hpi gets its own control instead of aborting as ambiguous. A lone control is reused across an arm's replicates, and denatured controls are resolved the same way.
• #16 - rf-count now counts only the references that have alignments, instead of running one samtools view per FASTA entry; whole-transcriptome runs (~250k human transcripts) previously never finished. RC files therefore list only covered references. The codon profile now gives the genome-route rf-count task the same resources as the transcriptome-route one.

Fixed

• #16 - Fuzzy control pairing only borrows a control from the same reference, so samples from different organisms can no longer pair.
• #16 - A sample_group/replicate group whose samples come from different organisms now stops with an error instead of pairing them.
• #16 - A sample_group/replicate group with more than one untreated or denatured sample now stops with a clear error instead of crashing; give runs of the same control the same sample name so they are concatenated.
• #16,#19 - rf-count no longer dies with "Unable to extract" when any of the first BAM records lacks an MD tag (an unmapped mate is enough), fixed upstream in RNA Framework 2.9.8 (#98).
• #16,#19 - rf-count no longer hangs at 100% CPU on references containing IUPAC ambiguity codes (e.g. Y, R); RNA Framework 2.9.8 reads them as N, so those positions report no reactivity (#99).
• #19 - The run summary in the MultiQC report now reads the samplesheet pH (or --pH), so DMS runs at pH ≥ 8 report ACGU reactive bases and, with four-base DMS, -nm 4 -4d; the tools always received the correct arguments.
• #16 - R2DT no longer fails on a missing .command.env when no sequences are extracted for drawing.
• #16 - --rfnorm_norm_method 1 (2-8% normalisation) is accepted again; the launch guard and schema rejected it while docs/usage.md documented it.
• #16 - The rf-fold flag letters quoted in the nextflow_schema.json descriptions of rffold_unconstrained, rffold_vienna_no_lonely_pairs, rffold_vienna_constrained, rffold_vienna_max_bp_span, rffold_fold_constraint_file and rffold_dotplot now match what the pipeline passes (-i, -nlp, -hc, -md, -c, -dp).
• #16 - .bp arc files list base pairs sorted by position instead of in rf-fold's arbitrary dot-plot order, so reruns give identical files.
• #16 - count/rfcount_summary_all_samples.tsv lists samples in name order instead of the order their rf-count tasks finished, so reruns give identical files.
• #16 - rf-normfactor no longer silently drops RC files whose sample name starts with a digit (e.g. 125ng_r1.rc): rf-rctools index numifies such a bare filename into an argument index, so the file is now passed with a directory prefix.

Removed

• #16 - --rffold_vienna_bp_span and --rffold_unpaired_constraint_file, which were declared but never used; rf-fold has no corresponding flags.

Dependencies

• #19 - RNA Framework 2.9.7 to 2.9.8

software	previously	now
RNA Framework	2.9.7	2.9.8

Deprecated

• #20 - --skip_markdup is replaced by --markdup (default false) and --fuzzy_untreated_pairing by --exact_untreated_pairing (default false), so boolean options default to false. The old names still work, inverted, with a deprecation warning.