Ian Morilla

Small signalling peptides (SSPs) coordinate development and stress response in plants, and the computational tools that identify them represent sequences using general-purpose protein language models trained across all known proteins. How much a representation learned on the SSP sequence space itself is worth, measured against generalist embeddings of varying size, has not to our knowledge been quantified. We pretrain PhytoPepBERT, a 169,567-parameter masked language model implemented from scratch in NumPy with no deep-learning frame-work, on 1,469 plant SSP precursors retrieved from UniProtKB spanning five peptide families (RALF, PSK, CLE, PEP, RGF) and a length-matched negative class of non-SSP plant proteins. Sequences are clustered at 70% alignment identity and every cross-validation fold is grouped by cluster, so no homologue of a held-out sequence appears in training. Domain pretraining is worth a great deal relative to where it starts. Family classification by k -nearest neighbours over frozen embeddings reaches 85.2% ± 2.4% against 58.0% ± 4.3% for an architecturally identical, identically initialised untrained backbone (chance 16.7%), and against 75.1% for dipeptide composition and 67.6% for BLOSUM62. The advantage is largest where the backbone stays frozen: linear probing gives 82.9% against 67.6% untrained, while full fine-tuning narrows the gap to 85.4% against 83.6%. Masked-residue accuracy is 20.5% against 3.4% untrained, elevated across all five peptide families and lowest on the non-SSP negatives. Benchmarked against ESM-2 at four scales under an identical protocol, those 169,567 domain-pretrained parameters buy approximately what 7.5M generalist parameters buy: PhytoPepBERT ties esm2_t6_8M_UR50D (86.1% ± 2.8%, two of five paired folds) and is overtaken thereafter, by 2.5 points at 35M, 7.0 at 150M and 11.6 at 650M, losing every paired fold above the smallest checkpoint. TAPE, the second representation the reference pipeline uses, leads by 3.6 points at 92M. Generalist scale therefore continues to pay on this task well beyond the point where a domain model of this size can follow. We report an exchange rate rather than a victory: domain pretraining delivers roughly a forty-fold parameter efficiency against the generalist curve, and nothing beyond it.

Whole-genome sequencing comprehensively captures coding, non-coding and structural variation in families with suspected inherited disorders, yet its clinical utility remains constrained by an interpretation bottleneck: selecting a handful of relevant variants from millions of candidates. Current rule-based pipelines, anchored in ACMG/AMP criteria, excel at identifying highly penetrant Mendelian alleles but frequently miss variants of low-to-moderate penetrance, non-coding alterations and germline-somatic interactions. Here we introduce PolyCLIP-T, a topology-guided multimodal framework that transforms variant selection from a classification problem into a geometric discovery task. By contrastively aligning DNA-sequence embeddings with functional annotations, PolyCLIP-T constructs a unified latent space in which the displacement between reference and alternate embeddings quantifies the molecular perturbation induced by each variant. Persistent homology then identifies stable topological components - coherent variant groups shared among affected relatives - that transcend single-variant scoring logic. Applied to six families with multi-morbid cancer, autoimmune and cardiovascular disease, PolyCLIP-T recovered non-coding and structural candidates overlooked by conventional pipelines and revealed pleiotropic networks spanning disease categories. This approach provides an interpretable, scalable solution for genome-first investigations of disorders driven by polygenic architectures that evade single-variant analysis. The framework was developed and benchmarked on deeply characterised familial cohorts selected for transgenerational multimorbidity; validation in larger, independent populations will be essential to establish its generalisability. An interactive web tool is freely available at https://www.polyclip-t.uma.es/.

Rapid Alkalinization Factor (RALF) peptides regulate plant growth and cell wall signaling, but their roles in fruit development remain unclear. Here, we characterized tomato (Solanum lycopersicum) fruit-associated RALF peptides and their interactions with leucine-rich repeat extensins (LRXs). Expression analyses identified SlRALF5, SlRALF7, and SlRALF10 as the main fruit-expressed RALFs. SlRALF10 was associated with early fruit development, whereas SlRALF5 and SlRALF7 remained expressed during ripening. Sequence analyses showed that SlRALF5/7 retain conserved motifs of canonical RALFs, while SlRALF10 displays divergent structural features and altered charge distribution. Synthetic SlRALF5 and SlRALF7 inhibited root growth and induced extracellular alkalinization, whereas SlRALF10 lacked both activities. Co-immunoprecipitation assays showed that all three peptides interact with the fruit-expressed proteins SlLRX2 and SlLRX5. Structural modeling predicted distinct electrostatic properties for the SlLRX5/SlRALF10 complex compared with SlRALF5. These results reveal structural and functional specialization among tomato fruit RALF peptides and suggest that distinct SlRALFs may differentially respond to cell wall remodeling during fruit development and ripening.

Lung transplantation remains the only definitive treatment for end-stage respiratory failure; however, it has substantial post-operative mortality risk. Current methods like the Lung Transplant Risk Index offer limited predictive performance. This study introduces a novel topological feature engineering model to assess mortality risk. The objective is to improve predictive accuracy by capturing complex temporal patterns while ensuring interpretability. A retrospective cohort study was conducted using clinical data from lung transplant recipients. The model integrates static and time-dependent variables through topological feature extraction, enabling sequential risk updating at transplantation, ICU admission, and throughout early post-operative course. Performance was compared to established methods using a held-out test set. Metrics included accuracy, sensitivity, specificity, and AUC. Interpretability was assessed using Shapley Additive explanations. The proposed model demonstrated superior predictive performance compared to traditional clinical risk scores (LTRI, CCI) and standard machine learning models. On the test dataset, it achieved 87.4% accuracy, 84.1% sensitivity, and 89.6% specificity, with an absolute AUC gain of 0.08 over the best non-topological baseline (p < 0.001). The model consistently outperformed existing approaches across subgroups including age, underlying disease, and transplant type. Shapley analysis revealed that dynamic variables such as early post-operative oxygenation trends, immunosuppressive load, and inflammatory markers were among the most critical contributors to mortality risk. The integration of topological features significantly enhances prediction of post-transplant mortality risk. These findings highlight topological transformers as a valuable tool for precision medicine and clinical decision support.

Plants integrate complex environmental signals through interconnected molecular networks, yet the fundamental rules governing this sensory integration remain unknown. Studying tomato roots interacting with fungal symbionts, we discovered how microbial effectors systematically reprogram plant sensory systems by coordinating transcriptional, metabolic, and phenotypic responses. Our multimodal analysis not only confirmed prior experimental findings through purely computational means, but also revealed novel integration hubs where sensory pathways converge. This dual validation approach revealed two key mechanisms: first, the rewiring of iron homeostasis through citrate-mediated redox control, and second, the targeted suppression of jasmonate defences. Furthermore, we demonstrate how nuclear splicing programs are isolated from metabolic noise. These findings establish a new paradigm for understanding plant-microbe communication by showing how symbionts exploit latent hubs where sensory pathways converge. The discovered integration logic provides both fundamental insights into plant perception and concrete targets for engineering stress-resilient crops.

Small signalling peptides (SSPs) serve as crucial mediators of cell‐to‐cell communication in plants, orchestrating diverse physiological processes from development to stress responses. While recent advances in sequencing technologies have improved genome annotation, the identification of novel SSPs remains challenging due to their small size, sequence diversity, and often transient expression patterns. To address this bottleneck, we developed S ² ‐PepAnalyst, a machine learning‐powered web tool that integrates plant‐specific datasets with advanced computational approaches for SSP prediction and classification. Our platform combines protein language models with geometric‐topological feature analysis to capture both sequence and structural characteristics of known SSP families. When validated against experimentally confirmed peptides, S ² ‐PepAnalyst achieved high predictive accuracy (99.5%) while maintaining low false‐negative rates. The tool successfully classified peptides into functionally distinct families (e.g., CLE, RALF) and identified non‐canonical SSPs that lack traditional signal peptides. Importantly, S ² ‐PepAnalyst demonstrated robust performance across both model plants and agriculturally important species. As a freely available resource ( https://www.s2‐pepanalyst.uma.es ), this tool will empower plant biologists to systematically explore the largely untapped repertoire of plant SSPs, facilitating discoveries in plant cell signalling and potential applications in crop improvement.

Ascorbate is the most abundant water-soluble antioxidant in plants and an essential molecule for normal plant development. Although present in all green plants, ascorbate concentrations vary among plant species and tissues. While ascorbate accumulation is a trait of nutritional, and therefore, agronomical interest, the impact of different concentrations on cellular homeostasis remains elusive. To shed light on this question, we compared Arabidopsis (Arabidopsis thaliana) lines with very low (vtc2 mutant, 20% of wild-type (WT) levels), low (vtc4 mutant, 65% of WT levels), and high (vtc2/OE-VTC2, 165% of WT levels) ascorbate concentration in 4-wk-old rosette leaves. An 80% reduction of ascorbate increased the expression of genes implicated in defense against pathogens but repressed genes associated with abiotic stress responses. Unexpectedly, lines with increased (165% of WT) and decreased (65% of WT) ascorbate levels shared 85% of induced transcription factors and the gene ontology terms associated with their transcriptional programs. We identified TRYPTOPHAN AMINOTRANSFERASE OF ARABIDOPSIS 1 (TAA1), a gene encoding the enzyme that catalyzes the first step of auxin biosynthesis, among the group of genes whose expression was positively correlated with ascorbate content. Using a combination of genetic and pharmacological approaches in fluorescent and histochemical reporter lines for auxin biosynthesis and signaling activity, we revealed that TAA1- and TAA1 RELATED 2 (TAR2)-mediated auxin biosynthesis is necessary for plants to cope with increased ascorbate concentration in a light-dependent manner, revealing a layer of complexity in the regulatory landscape of redox homeostasis.

Background and Objective Lung transplantation remains the only definitive treatment for patients with end-stage respiratory failure; however, it is burdened by a substantial risk of post-operative mortality. Current risk stratification methods, such as the Lung Transplant Risk Index, offer limited predictive performance and interpretability. This study introduces a novel predictive model based on topological transformers to assess mortality risk following lung transplantation. The objective is to improve predictive accuracy by capturing complex temporal patterns in clinical data while ensuring model interpretability to inform clinical decisions. Methods A retrospective cohort study was conducted using clinical data from lung transplant recipients. The model integrates both static and time-dependent clinical variables through a transformer-based architecture that incorporates topological features derived from patients’ temporal trajectories. Model performance was compared to established methods using a held-out test set. The evaluation metrics included accuracy, sensitivity, specificity, and the area under the receiver operating characteristic curve. Model interpretability was assessed using Shapley Additive explanations to identify and rank the most influential predictors of mortality. Results The proposed model demonstrated superior predictive performance compared to the Lung Transplant Risk Index and other benchmark models. On the test dataset, it achieved an accuracy of 87.4%, sensitivity of 84.1%, and specificity of 89.6%. The model consistently outperformed existing approaches across different subgroups, including age, underlying disease, and transplant type. Shapley-based interpretability analysis revealed that dynamic variables such as early post-operative oxygenation trends, immunosuppressive load, and inflammatory markers were among the most critical contributors to mortality risk. Conclusions The integration of topological features within a transformer-based framework significantly enhances the prediction of post-transplant mortality risk. By offering both improved predictive power and model transparency, this approach supports more precise and personalised risk stratification in lung transplantation. These findings highlight the potential of topological transformers as a valuable tool in the broader context of precision medicine and clinical decision support.