Fix missing Remotes field in DESCRIPTION so that pak::pkg_install() can resolve the GitHub-only dependency MiscMetabar when installing tidypq standalone.
tidypq 0.2.0
Breaking changes
Contamination handling is unified around a detector then filter design: each identify_contam_*_pq() detector returns a contam_tbl (one row per flagged taxon), and the single verb filter_contam_pq() removes the flagged taxa. The previous one-shot functions are removed with no deprecation, so old scripts fail with could not find function. Update calls as follows: contam_corr_pq() to identify_contam_corr_pq(); contam_blocklist_pq() to identify_contam_blocklist_pq(); identify_taxa_primer_pq() and filter_taxa_primer_pq() to identify_contam_primer_pq(); chimera_removal_dada2() and chimera_removal_vs_ref() to identify_contam_chimera_pq(method = "dada2") and identify_contam_chimera_pq(method = "vsearch_ref"); neg_control_classify_pq(), neg_control_clean_pq(), and neg_control_diag_pq() to identify_contam_negcontrol_pq(). In every case, pass the result to filter_contam_pq() to remove taxa.
The rich diagnostic plots formerly bundled in detector return values are removed; a lightweight plot() / autoplot() method on contam_tbl replaces them, with publication-grade figures deferred to the ggplotpq package.
decontam_sam_control() and decontam_taxa_control() are renamed decontam_control_samples_pq() and decontam_control_taxa_pq() (old names fail with could not find function); their behaviour is unchanged. They remain control-based per-occurrence cleaners and are deliberately outside the contam_tbl detector family.
New features
uncross2_pq() can now estimate the rate of tax-switching using internal function estimate_xtalk_rate_pq(). It estimates the cross-talk (tag-jump) rate f de novo from an OTU table following Edgar 2018 (UNCROSS2, eqs. 1-2), with rate = "edgar2018" (paper eq. 2) or rate = "fraction" (the raw low-read fraction that usearch reports), returning NA when the estimate is undetermined.
filter_contam_pq() removes the taxa flagged in a contam_tbl (from one detector, or several combined with rbind()), returning the phyloseq object unchanged when nothing is flagged.
merge_markers_pq() combines several phyloseq objects sequenced with different markers (e.g. 16S and ITS) into a single object: it full-joins the sam_data, disambiguates taxa names by their marker so taxa from different objects are never merged, records the marker of origin in a new tax_table column, and zero-fills unobserved marker/sample occurrences.
tax_table_to_df() converts a phyloseq tax_table into a tibble with numeric and list columns, lifting the character-only limitation of the tax_table matrix: numeric-looking ranks are converted to their natural type via convert, and columns named in split become genuine list columns.
identify_contam_blocklist_pq() flags taxa whose genus matches a curated blocklist of known reagent and laboratory contaminants (Salter et al. 2014, micRoclean, GRIMER); known_contaminant_genera() returns the default blocklist.
identify_contam_chimera_pq() flags chimeric taxa using dada2 de novo detection (method = "dada2") or vsearch reference-based UCHIME (method = "vsearch_ref").
identify_contam_corr_pq() flags taxa whose relative abundance correlates negatively with sample depth (GRIMER-inspired), with cor_abs, cor_rel, slope, and p_value evidence columns.
identify_contam_negcontrol_pq() flags taxa from negative-control occurrence patterns and records a subtype of artifact, lab_contaminant, or sample_contaminant; only artifact and lab_contaminant are flagged by default.
identify_contam_primer_pq() flags taxa whose representative sequence embeds a primer (forward or reverse complement) using IUPAC-aware matching via Biostrings.
uncross2_pq() removes per-cell cross-talk (tag-jumps) using Edgar’s published UNCROSS2 score (eq. 3), with a fixed rate f or f = "auto" for the de novo estimate, method = "cut" or "subtract", and an optional return_scores for the score and flag matrices; it is a faithful, gentler implementation of the paper and does not reproduce the more aggressive usearch binary filtering.
Every verb that returns a phyloseq object (filter_*_pq(), select_*_pq(), mutate_*_pq(), slice_*_pq(), arrange_*_pq(), rename_*_pq(), filter_occurrences_pq(), filter_contam_pq(), filter_chimera_pq(), decontam_control_samples_pq(), decontam_control_taxa_pq()) now reorders a desynchronised sam_data or tax_table back to the canonical otu_table order (emitting a message when a reorder was needed) and verifies the result with MiscMetabar::verify_pq(check_order = TRUE) before returning. This guarantees the returned object keeps sample_names()/taxa_names() aligned across slots, fixing a class of silent positional bugs (e.g. sam_data columns no longer ending up attached to the wrong samples).
tidypq 0.1.1
chimera_removal_vs_ref() removes chimeric sequences using UCHIME reference-based detection via vsearch.
contam_blocklist_pq() flags taxa whose genus belongs to a curated blocklist of known reagent and laboratory contaminants (compiled from Salter et al. 2014, micRoclean and GRIMER), a taxonomy-only method requiring no negative controls; the accompanying known_contaminant_genera() returns the default blocklist for inspection or extension.
contam_corr_pq() detects potential contaminant taxa by examining the correlation between each taxon’s relative abundance and total sample reads. Contaminants show a negative correlation (higher proportion in low-biomass samples). Returns per-taxon statistics (cor_abs, cor_rel, slope, p_value), a flagged contaminants subset, and a three-panel diagnostic plot. Inspired by the GRIMER tool.
neg_control_classify_pq() classifies taxa found in negative controls into artifact, lab_contaminant and sample_contaminant categories based on read abundance and occurrence patterns, returning the per-taxon classification together with summary, distribution, heatmap and ordination plots.
