Biology subjects

Parada, P.

Publications and source records attributed to Parada, P..

5 recordsLinked to original sources

bioRxiv2026

Pretrained deep-learning ITS classifiers read the flanking regions, not the ITS2 barcode, and so fail on the amplicon that environmental fungal surveys sequence

Background. Deep-learning classifiers for the fungal internal transcribed spacer (ITS) report accuracies above 90% and are increasingly proposed for environmental metabarcoding. That application differs from the benchmarks in two ways: surveys sequence a primer- bounded subregion, most often ITS2, rather than the full-length reference sequences the models were trained and tested on, and much of what they recover belongs to genera absent from any reference. We tested whether the reported accuracies transfer. Design. Five methods were scored on the same 5,222 queries, one sequence per genus, at two loci: the full-length ITS record, and the ITS2 subregion of that identical record. The methods are the two pretrained deep-learning classifiers distributed with MycoAI, a convolutional network and a transformer sharing a training corpus of 5.23M sequences; two reference-based methods, best-hit alignment and SINTAX; and HiTaC, a hierarchical logistic regression over k-mer counts, which we fitted ourselves to the reference the other two consult. Queries were stratified by whether the query's genus lies in the pretrained models' label space, recovered from the distributed models themselves. A novel genus is one outside that label space, and outside the reference by the same rule, so no method here can return its correct name; seen genera are the rest (2.2). Results. The design favours the pretrained models, whose queries come from the public dataset they were trained on. Even so, on full-length ITS the other three methods exceed both of them at every rank and in both strata, best-hit alignment recovering the correct family for 92.3% of seen-genus queries against 77.9% and 76.5%. Restricting the identical records to ITS2 costs the reference-based methods under four percentage points of seen-genus family accuracy and costs the two pretrained models 49.6 and 58.0, reducing them to 28.3% and 18.5%; HiTaC refitted at that locus recovers 89.2%, so neither learned classification nor the amplicon is what fails. An ablation identifies the cause. Grafting each query's unaltered ITS2 between the flanking regions of a donor record from a different phylum returns the donor's family for 34.4% of queries in the convolutional model and 63.7% in the transformer, against the query's own for 4.2% and 0.8%, from a baseline near zero where no donor is present. The predictions therefore follow the flanking regions rather than the ITS2 barcode, which accounts for the collapse and predicts the same failure for any subregion amplicon. Compounding this, on ITS2 the classifiers' confidence score all but ceases to separate novel from seen genera (AUROC 0.541 and 0.503, the latter at chance, against 0.866 for alignment identity), and the convolutional model is in addition substantially overconfident there, so for that model the failure is not detectable from its own output at all. Recommendation. Reported accuracies for such models should specify the amplicon region of the evaluation, state the length distribution of the training corpus, and be accompanied by a same-query classical baseline.

bioinformatics↗

bioRxiv2026

A calibrated novelty flag for fungal ITS metabarcoding: choosing the error rate at which sequences are declared new

O_LIEnvironmental fungal surveys routinely recover internal transcribed spacer (ITS) sequences that cannot be assigned at fine taxonomic ranks, the so-called fungal "dark matter." Such sequences are set aside by thresholding a similarity or confidence score at a conventional value. Those conventions do abstain, but the error rate a threshold implies is neither stated nor selectable, and a threshold defined for one kind of score does not transfer to another. C_LIO_LIWe present a conformal novelty flag that supplies what is missing: each query receives a p-value with a distribution-free guarantee that the rate of falsely declaring a known sequence novel is bounded by a user-chosen . We evaluate it on a leave-one-genus-out benchmark built from UNITE, on alignment identity, a k-mer bootstrap consensus and two neural classifiers output probabilities, and on a soil fungal dataset. C_LIO_LIThe flag holds its nominal rate across two orders of magnitude in , so an operating point can be chosen rather than inherited: at = 0.05 it fires on 4.3% of known-genus queries and recovers 19.2% of genuinely novel genera. The cutoff holding a 5% error rate here is 64.8% identity, nowhere near the customary 97%, showing how little a threshold carries its error rate between datasets. Coverage transferred across eleven settings spanning those four scores, two amplicon regions and a sevenfold change in reference size, all within 1.1 percentage points of nominal, while detection ranged from 5.2% to 53.3%: the guarantee is on the error rate and not on power, and two of our settings are valid but uninformative. Applied to soil data the flag identifies 20.7% of amplicon sequence variants as novel at a controlled 5% error rate, 16.3% under an abundance filter. Half of those recur near-identically among GlobalFungis unnamed environmental variants while fewer than one in ten matches a named species hypothesis, a sixfold skew towards the uncatalogued against 1.9-fold for sequences the flag passes. C_LIO_LIThe flag turns an arbitrary cutoff into a decision with a stated error rate, and in doing so converts dark matter from a residue into a set of prioritizable targets for formal description. C_LI

bioinformatics↗

bioRxiv2026

SSUplex: fast, both-strand extraction and origin-sorting of small-subunit rRNA for environmental DNA metabarcoding

Ribosomal RNA metabarcoding sits at the center of how we characterize microbial and eukaryotic communities in environmental samples, and long-read sequencing has made full-length small-subunit (SSU; 16S/18S) profiling routine. The broadly conserved primers that make rRNA such a convenient marker are also its liability: by design they co-amplify organellar (mitochondrial, chloroplast) and cross-domain SSU alongside the intended target. Left unsorted before taxonomic assignment, these passengers are systematically misclassified, and the error propagates straight into estimates of community composition and diversity. Reads must therefore be detected, extracted, and sorted by origin before they ever reach a classifier. We present SSUplex, an open-source tool that detects SSU rRNA, assigns each read to one of five origins (bacteria, archaea, eukaryota, mitochondria, chloroplast), and extracts the SSU region for downstream classification. SSUplex reimplements the extraction-and-origin logic of the widely used Metaxa2 in the Rust programming language, scans both strands, and ships as a single dependency-light binary suited to long-read (Oxford Nanopore, PacBio HiFi) and short-read data. Benchmarked against Metaxa2 on public data, SSUplex reproduces Metaxa2 origin calls on full-length reads (96.8% concordance) and matches its extraction speed on small inputs, then pulls away to run up to [~]3.4x faster with [~]35% lower peak memory at 200,000 reads, the per-sample scale a long-read amplicon run typically reaches. We are candid about a genuine, measured trade-off in the origin-ranking statistic, and we pinpoint the bacteria-versus-mitochondria boundary as the methods one intrinsically lower-confidence edge. For the now-common workflow in which origin-sorted reads are handed to a dedicated classifier rather than classified in place, SSUplex is a fast, reproducible, embeddable stand-in for Metaxa2s extraction role. Source code and a benchmark harness that regenerates every result from public data are available under the MIT license at https://github.com/ayobi/ssuplex.

bioinformatics↗

bioRxiv2026

EMITS: expectation-maximization abundance estimation for fungal ITS communities from long-read sequencing

As long-read amplicon sequencing becomes routine for fungal metabarcoding, species-level abundance estimation from ITS amplicons remains limited by naive best-hit classification, which misattributes reads among closely related species sharing similar ITS sequences and fragments abundance across redundant database entries. Expectation-maximization (EM) approaches developed for full-length 16S rRNA, notably EMU [Curry et al., 2022], have recently been benchmarked for fungal ITS metabarcoding [Graetz et al., 2025], but applying EMU to ITS requires custom reference database construction and uses parameters originally tuned for 16S. Here we present EMITS, a Rust-based tool that applies EM to iteratively resolve ambiguous read-to-reference mappings from minimap2 alignments against the UNITE database, producing probabilistic species-level abundance estimates. EMITS provides UNITE-native header parsing with automatic accession aggregation, empirically tuned platform presets for current Oxford Nanopore (R10.4.1, R9.4.1, Duplex) and PacBio HiFi chemistries, and integration with ITSxRust [OBrien et al., 2026] for upstream ITS region extraction. We validated EMITS using three complementary approaches and benchmarked it against both naive best-hit counting and EMU (with a UNITE-formatted reference database). In controlled simulations, EM reduced L1 error by 80- 92% compared to naive counting under realistic noise conditions. On the ATCC fungal ITS mock community, EMITS provided superior within-genus species resolution in taxonomically challenging genera: it correctly identified Trichophyton mentagrophytes (2.21%) where EMU misattributed substantial abundance to T. tonsurans (1.54%); it suppressed Penicillium rubens false positives (0.002% vs. EMU 0.58%); and it more accurately consolidated Nakaseomyces glabratus abundance across UNITE accessions (12.40% vs. EMU 9.95%). On a 21-species synthetic UNITE community lacking substantial within-genus difficulty, all three methods detected expected species at 100% sensitivity, with aggregate L1 errors of 8.64% (naive), 7.48% (EMITS), and 6.71% (EMU). Together with ITSxRust for upstream ITS extraction, EMITS provides a complete pipeline tuned for long-read fungal amplicon profiling.

bioinformatics↗

bioRxiv2026

ITSxRust: ITS region extraction with partial-chain recovery and structured diagnostics for long-read amplicon sequencing

As long-read amplicon sequencing (e.g., Oxford Nanopore and PacBio HiFi) becomes routine for fungal metabarcoding, identifying and extracting ITS subregions at scale has become a throughput and robustness bottleneck. The nuclear ribosomal internal transcribed spacer (ITS) region is the formal DNA barcode for fungi and is widely used for taxonomic profiling of fungal communities [Schoch et al., 2012]. Standard preprocessing locates conserved ribosomal flanks with hidden Markov profile models (profile-HMMs) to extract ITS1, 5.8S, ITS2, or the full ITS, as implemented in ITSx [Bengtsson-Palme et al., 2013] and ITSxpress [Rivers et al., 2018, Einarsson and Rivers, 2024]. Here we describe ITSxRust, a Rust-based ITS extractor designed for long-read scale. ITSxRust coordinates HMMER searches with efficient Rust-native I/O and sequence processing, optionally reduces redundant searches via dereplication, provides ONT and HiFi parameter presets, and emits structured failure diagnostics and QC summaries. On an Oxford Nanopore ITS dataset (54,659 reads), ITSxRust extracted the full ITS region from 75.3% of reads, exceeding both ITSx (69.9%) and ITSxpress v2 (41.4%), while running 4.6x faster than ITSx. In addition, a partial-chain fallback strategy that extracts subregions using two-anchor pairs when the full four-anchor chain is unavailable recovered an additional 10,725 reads that would otherwise be discarded.

bioinformatics↗