Exploring the fungal circular RNA (circRNA) landscape is bottlenecked by both experimental and computational limits. While standard mRNA-seq systematically discards circRNAs due to their lack of poly(A) tails, high-cost total RNA-seq remains prohibitive for large-scale screening. Consequently, discovery relies heavily on computational prediction. However, existing models trained exclusively on human or plant sequences fail in fungi because of the extreme genomic divergence across fungal lineages, which span from intron-poor Candida to intron-rich filamentous fungi. As a result, no computational framework currently exists for de novo fungal circRNA prediction, leaving the vast majority of non-model fungi entirely inaccessible. Results: We present mycoCirc, the first end-to-end pan-fungal multi-modal pretrained model for fungal circRNA prediction, integrating three mandatory modalities with bidirectional cross-attention for donor-acceptor site interaction. Pre-trained on 22 strains with 16,483 positive gene-circRNA associations spanning Ascomycete yeast, Basidiomycete yeast, and Filamentous fungi groups and fine-tuned per group using 5-fold cross-validation, mycoCirc achieved AUROC 0.69-0.70 on held-out test species under Mode A (Genome+GTF, no RNA-seq), substantially outperforming JEDI (0.51-0.57) and CircPCBL (0.49-0.53). Cross-species evaluation on four independent fungi datasets demonstrated robust generalization across all fine-tuned variants (AUROC 0.63-0.72). Beyond gene-level classification, the JunctionEncoder module enabled backsplicing junction identification for detailed circRNA validation. We further build mycoCircAtlas, a companion database providing 319,860 high-confidence gene-circRNA predictions across 768 fungal species from Ensembl Fungi Release 113, enabling researchers to query precomputed predictions and design validation primers without local model deployment.

Phaeobacter inhibens T5T harbors three LuxI/LuxR quorum-sensing systems: one encoded by the host and one in each of two prophages. Each prophage quorum-sensing autoinducer activates its own lysogeny program and that of the co-resident prophage. By driving competitors toward lysogeny, prophages deny their rivals access to susceptible hosts. One prophage engages in a bidirectional interaction with the host: the prophage activates host quorum sensing, while the host represses prophage quorum sensing. Because the host and prophage autoinducers exert opposing effects on the same prophage promoter, the prophage's lysis-lysogeny decision depends on the ratio of the two autoinducers rather than their absolute concentrations. Ratiometric sensing allows the prophage to infer the relative abundances of infected and uninfected hosts and to commit to lysis only when uninfected hosts predominate. Phage quorum-sensing modules are widespread and undergo diversification and exchange. These findings reveal how prophages surveil and manipulate competitors that share chemical environments.

Nanopore direct RNA sequencing (DRS) has transformed transcriptomics by enabling single-molecule, long-read sequencing of native RNA without the need for reverse transcription or amplification. In contrast to short-read RNA-seq and cDNA-based long-read approaches, DRS can simultaneously capture multiple RNA modifications, full-length transcript architecture, alternative splicing patterns, and poly(A) tail features within individual molecules, thereby providing an integrated view of transcriptomic and epitranscriptomic regulation. In this comprehensive review, we outline the biophysical principles underlying nanopore DRS and trace its technological evolution. We compare its performance with short-read RNA sequencing, long-read cDNA sequencing, and conventional RNA-modification mapping strategies, highlighting its advantages in isoform-resolved quantification and multilayer RNA feature integration, while also clarifying contexts in which alternative or combined approaches may be more appropriate for robust biological interpretation. We further summarize optimized experimental workflows, including library construction strategies tailored to diverse RNA biotypes (mRNA, rRNA, tRNA, circRNA, miRNA, and nonpoly(A) transcripts), as well as recommended quality-control procedures and sequencing optimization practices. Emphasizing recent computational advances and translational applications of DRS, we cover state-of-the-art algorithms for RNA modification detection, transcript reconstruction, and isoform quantification. We also propose analytical pipelines for poly(A) tail length inference and integrative frameworks that jointly analyze these regulatory layers. We distinguish direct nanopore signals from computational inferences to define confidence levels and emphasize benchmarking and orthogonal validation of readouts. Practical implementation examples are included to facilitate reproducible analysis. Finally, we highlight emerging applications of integrated DRS, including the resolution of complex transcriptomes, the characterization of coordinated epitranscriptomic regulation, and the identification of disease-associated RNA signatures. We also discuss current technical challenges and future perspectives, particularly in relation to multi-omics integration and the broader deployment of DRS in precision medicine as well as in plant and animal research.

CRISPR-Cas9 genome-editing efficiency is strongly influenced by the sequence composition and positional context of single-guide RNAs (sgRNAs). Although numerous deep learning–based models have been developed to predict Cas9 efficiency from sgRNA sequences, most operate as black boxes, offering limited insight into the sequence determinants underlying Cas9 activity. In addition, previous studies often overlook how the positional context of sequence motifs within sgRNAs influences their effects on Cas9 binding or cleavage.  We introduce DeepCC9, an interpretable machine learning framework that combines explicit sequence feature extraction with a residual block–based deep architecture to improve interpretability and identify composition- and position-based motifs governing Cas9 genome-editing efficiency. We applied this method to multiple Cas9 variant datasets, achieving superior predictive performance compared with existing methods while enabling direct interpretation of sequence motifs and their positional effects. Our analysis uncovered 74 sequence motifs enriched or depleted at specific positions within sgRNAs and strongly associated with Cas9 efficiency, providing mechanistic insight into sequence features that influence guide performance. Together, these results establish DeepCC9 as a generalizable and interpretable framework for modeling sequence–function relationships and advancing the understanding of the sequence determinants underlying CRISPR-Cas9 genome editing.

Antimicrobial peptide discovery is constrained less by the number of molecules that can be generated than by the choice of which few should be tested against a defined pathogen. Most peptide generators produce broadly antimicrobial-like sequences and leave target specificity to downstream filters. Here we show that pathogen context can be introduced during generation. AMPHORA conditions a peptide-native generator on target class, pathogen genome features and strain-description text. Matched, ablated and shuffled controls showed that aligned pathogen inputs redirected generated libraries beyond coarse activity labels, whereas global shuffling weakened this effect. Same-noise counterfactuals showed that strain descriptions drove larger sequence changes, whereas genome features more strongly affected predicted structural properties. Species-level analyses revealed target-dependent enrichment. Matched bacterial inputs also shifted APEX-predicted activity rankings relative to class-only generation. The resulting libraries remained diverse, largely non-memorizing and compatible with predicted peptide-like structural features. Together, these results establish pathogen-context conditioning as a new paradigm for computational library reshaping in antimicrobial peptide generation.

Phage therapy offers a promising solution to the antimicrobial resistance crisis. However, a major concern preventing the adoption of phage therapy is the potential for unintended consequences of phage release; both in regard to preventing the spread of phage resistance, and the proliferation of a non-endemic virus into the microbial ecosystem. Conditional replication (biocontainment) of phages through bioengineering may address these concerns, but the impact on bactericidal efficacy is unknown. Here, we created a biocontained T7 phage (T7Δcapsid) lacking the major structural capsid gene, gp10AB, that can only replicate on Escherichia coli strains expressing gp10AB in trans, and assessed its bactericidal efficacy compared with wild-type T7. Congruent with model predictions, T7Δcapsid was only able to clear a well-mixed culture of E. coli at a multiplicity of infection (MOI) of 10 or higher, whereas wild-type T7 prohibited growth at an MOI of 0.1. The reduction in efficacy was more evident in a complex structured environment within a microfluidic device, where phage success depends on its ability to penetrate a microbial niche via propagation. In this environment, T7Δcapsid was unable to propagate into the bacterial population and unlike wild-type T7, had no impact on the population's growth. This study shows that whilst biocontainment of phages may improve the biosafety of phage therapy, it comes at the cost of its propagation efficacy and niche penetration in relevant environments.