Explore the vast range of titles available.

Resistance to EGFR inhibitors (EGFRi) presents a major obstacle in treating non-small cell lung cancer (NSCLC). One of the most exciting new ways to find potential resistance markers involves running functional genetic screens, such as CRISPR, followed by manual triage of significantly enriched genes. This triage process to identify ‘high value’ hits resulting from the CRISPR screen involves manual curation that requires specialized knowledge and can take even experts several months to comprehensively complete. To find key drivers of resistance faster we build a recommendation system on top of a heterogeneous biomedical knowledge graph integrating pre-clinical, clinical, and literature evidence. The recommender system ranks genes based on trade-offs between diverse types of evidence linking them to potential mechanisms of EGFRi resistance. This unbiased approach identifies 57 resistance markers from >3,000 genes, reducing hit identification time from months to minutes. In addition to reproducing known resistance markers, our method identifies previously unexplored resistance mechanisms that we prospectively validate. Resistance to EGFR inhibitors presents a major obstacle in treating non-small cell lung cancer. Here, the authors develop a recommender system ranking genes based on trade-offs between diverse types of evidence linking them to potential mechanisms of EGFRi resistance.

Background Plant-pathogenic oomycetes are responsible for economically important losses in crops worldwide. Phytophthora palmivora , a tropical relative of the potato late blight pathogen, causes rotting diseases in many tropical crops including papaya, cocoa, oil palm, black pepper, rubber, coconut, durian, mango, cassava and citrus. Transcriptomics have helped to identify repertoires of host-translocated microbial effector proteins which counteract defenses and reprogram the host in support of infection. As such, these studies have helped in understanding how pathogens cause diseases. Despite the importance of P. palmivora diseases, genetic resources to allow for disease resistance breeding and identification of microbial effectors are scarce. Results We employed the model plant Nicotiana benthamiana to study the P. palmivora root infections at the cellular and molecular levels. Time-resolved dual transcriptomics revealed different pathogen and host transcriptome dynamics. De novo assembly of P. palmivora transcriptome and semi-automated prediction and annotation of the secretome enabled robust identification of conserved infection-promoting effectors. We show that one of them, REX3, suppresses plant secretion processes. In a survey for early transcriptionally activated plant genes we identified a N. benthamiana gene specifically induced at infected root tips that encodes a peptide with danger-associated molecular features. Conclusions These results constitute a major advance in our understanding of P. palmivora diseases and establish extensive resources for P. palmivora pathogenomics, effector-aided resistance breeding and the generation of induced resistance to Phytophthora root infections. Furthermore, our approach to find infection-relevant secreted genes is transferable to other pathogen-host interactions and not restricted to plants.

Background CRISPR ( C lustered R egularly I nterspaced S hort P alindromic R epeats) is a prokaryotic adaptive defence system that provides resistance against alien replicons such as viruses and plasmids. Spacers in a CRISPR cassette confer immunity against viruses and plasmids containing regions complementary to the spacers and hence they retain a footprint of interactions between prokaryotes and their viruses in individual strains and ecosystems. The human gut is a rich habitat populated by numerous microorganisms, but a large fraction of these are unculturable and little is known about them in general and their CRISPR systems in particular. Results We used human gut metagenomic data from three open projects in order to characterize the composition and dynamics of CRISPR cassettes in the human-associated microbiota. Applying available CRISPR-identification algorithms and a previously designed filtering procedure to the assembled human gut metagenomic contigs, we found 388 CRISPR cassettes, 373 of which had repeats not observed previously in complete genomes or other datasets. Only 171 of 3,545 identified spacers were coupled with protospacers from the human gut metagenomic contigs. The number of matches to GenBank sequences was negligible, providing protospacers for 26 spacers. Reconstruction of CRISPR cassettes allowed us to track the dynamics of spacer content. In agreement with other published observations we show that spacers shared by different cassettes (and hence likely older ones) tend to the trailer ends, whereas spacers with matches in the metagenomes are distributed unevenly across cassettes, demonstrating a preference to form clusters closer to the active end of a CRISPR cassette, adjacent to the leader, and hence suggesting dynamical interactions between prokaryotes and viruses in the human gut. Remarkably, spacers match protospacers in the metagenome of the same individual with frequency comparable to a random control, but may match protospacers from metagenomes of other individuals. Conclusions The analysis of assembled contigs is complementary to the approach based on the analysis of original reads and hence provides additional data about composition and evolution of CRISPR cassettes, revealing the dynamics of CRISPR-phage interactions in metagenomes.

The expansion of plants onto land was a formative event that brought forth profound changes to the earth’s geochemistry and biota. Filamentous eukaryotic microbes developed the ability to colonize plant tissues early during the evolution of land plants, as demonstrated by intimate, symbiosis-like associations in >400 million-year-old fossils. However, the degree to which filamentous microbes establish pathogenic interactions with early divergent land plants is unclear. Here, we demonstrate that the broad host-range oomycete pathogen Phytophthora palmivora colonizes liverworts, the earliest divergent land plant lineage. We show that P. palmivora establishes a complex tissue-specific interaction with Marchantia polymorpha, where it completes a full infection cycle within air chambers of the dorsal photosynthetic layer. Remarkably, P. palmivora invaginates M. polymorpha cells with haustoria-like structures that accumulate host cellular trafficking machinery and the membrane syntaxin MpSYP13B, but not the related MpSYP13A. Our results indicate that the intracellular accommodation of filamentous microbes is an ancient plant trait that is successfully exploited by pathogens like P. palmivora.

The leaf outer epidermal cell wall acts as a barrier against pathogen attack and desiccation, and as such is covered by a cuticle, composed of waxes and the polymer cutin. Cutin monomers are formed by the transfer of fatty acids to glycerol by glycerol-3-phosphate acyltransferases, which facilitate their transport to the surface. The extent to which cutin monomers affect leaf cell wall architecture and barrier properties is not known. We report a dual functionality of pathogen-inducible GLYCEROL-3-PHOSPHATE ACYLTRANSFERASE 6 (GPAT6) in controlling pathogen entry and cell wall properties affecting dehydration in leaves. Silencing of Nicotiana benthamiana NbGPAT6a increased leaf susceptibility to infection by the oomycetes Phytophthora infestans and Phytophthora palmivora, whereas overexpression of NbGPAT6a-GFP rendered leaves more resistant. A loss-of-function mutation in tomato SlGPAT6 similarly resulted in increased susceptibility of leaves to Phytophthora infection, concomitant with changes in haustoria morphology. Modulation of GPAT6 expression altered the outer wall diameter of leaf epidermal cells. Moreover, we observed that tomato gpat6-a mutants had an impaired cell wall–cuticle continuum and fewer stomata, but showed increased water loss. This study highlights a hitherto unknown role for GPAT6-generated cutin monomers in influencing epidermal cell properties that are integral to leaf–microbe interactions and in limiting dehydration.

Most machine learning approaches to scientific discovery frame hypotheses as end-to-end predictions, obscuring the incremental structure of scientific reasoning. We propose The Hypothesis Game, a symbolic formalism for hypothesis refinement in which LLM agents operate on a shared hypothesis state using a fixed grammar of reasoning moves. The framework is motivated by the observation that scientific progress often proceeds through small, localized revisions, grounded in domain context, rather than extensive rewrites. We instantiate a minimal game with LLM agents and evaluate it on pathway-level mechanistic refinement tasks. In the primary setting of corruption recovery, where hypotheses contain controlled errors, the game-based approach consistently removes more errors and achieves higher precision than strong prompting baselines, while preserving valid structure through incremental edits. In a secondary reconstruction setting from partial cues, it performs comparably to the strongest baseline, indicating that explicit move-based refinement remains competitive even when ground-truth recovery is difficult. These findings support game-based reasoning as a principled route to more controllable, interpretable, and transferable hypothesis refinement systems for scientific discovery.

We propose the molecular omics network (MOOMIN) a multimodal graph neural network used by AstraZeneca oncologists to predict the synergy of drug combinations for cancer treatment. Our model learns drug representations at multiple scales based on a drug-protein interaction network and metadata. Structural properties of compounds and proteins are encoded to create vertex features for a message-passing scheme that operates on the bipartite interaction graph. Propagated messages form multi-resolution drug representations which we utilized to create drug pair descriptors. By conditioning the drug combination representations on the cancer cell type we define a synergy scoring function that can inductively score unseen pairs of drugs. Experimental results on the synergy scoring task demonstrate that MOOMIN outperforms state-of-the-art graph fingerprinting, proximity preserving node embedding, and existing deep learning approaches. Further results establish that the predictive performance of our model is robust to hyperparameter changes. We demonstrate that the model makes high-quality predictions over a wide range of cancer cell line tissues, out-of-sample predictions can be validated with external synergy databases, and that the proposed model is data efficient at learning.

Duplication of nodes is a common problem encountered when building knowledge graphs (KGs) from heterogeneous datasets, where it is crucial to be able to merge nodes having the same meaning. OntoMerger is a Python ontology integration library whose functionality is to deduplicate KG nodes. Our approach takes a set of KG nodes, mappings and disconnected hierarchies and generates a set of merged nodes together with a connected hierarchy. In addition, the library provides analytic and data testing functionalities that can be used to fine-tune the inputs, further reducing duplication, and to increase connectivity of the output graph. OntoMerger can be applied to a wide variety of ontologies and KGs. In this paper we introduce OntoMerger and illustrate its functionality on a real-world biomedical KG.

The leaf epidermal wall is covered by a cuticle, composed of cutin and waxes, which protects against dehydration and constitutes a barrier against pathogen attack. Cutin monomers are formed by the transfer of 16- or 18-carbon fatty acids to glycerol by glycerol-3-phosphate acyltransferase (GPAT) enzymes, which facilitates their transport to the plant surface. Here we address the dual functionality of pathogen-inducible Glycerol phosphate acyltransferase 6 (GPAT6) in controlling pathogen entry and dehydration in leaves. Silencing of Nicotiana benthamiana NbGPAT6a increased leaf susceptibility to the oomycetes Phytophthora infestans and P. palmivora, whereas stable overexpression of NbGPAT6a-GFP rendered leaves more resistant to infection. A loss-of-function mutation of the orthologous gene in tomato (Solanum lycopersicum), SlGPAT6, similarly resulted in increased susceptibility of leaves to Phytophthora infection concomitant with altered intracellular infection structure morphology. Conversely, Botrytis cinerea disease symptoms were reduced. Modulation of GPAT6 expression predominantly altered the outer cell wall of leaf epidermal cells. The impaired cell wall-cuticle continuum of tomato gpat6-a mutants resulted in increased water loss and these plants had fewer stomata. Our work highlights a hitherto unknown role for GPAT6-generated cutin monomers in controlling epidermal cell properties that are integral to leaf-microbe interactions and limit dehydration.

In this paper, we introduce ChemicalX, a PyTorch-based deep learning library designed for providing a range of state of the art models to solve the drug pair scoring task. The primary objective of the library is to make deep drug pair scoring models accessible to machine learning researchers and practitioners in a streamlined framework.The design of ChemicalX reuses existing high level model training utilities, geometric deep learning, and deep chemistry layers from the PyTorch ecosystem. Our system provides neural network layers, custom pair scoring architectures, data loaders, and batch iterators for end users. We showcase these features with example code snippets and case studies to highlight the characteristics of ChemicalX. A range of experiments on real world drug-drug interaction, polypharmacy side effect, and combination synergy prediction tasks demonstrate that the models available in ChemicalX are effective at solving the pair scoring task. Finally, we show that ChemicalX could be used to train and score machine learning models on large drug pair datasets with hundreds of thousands of compounds on commodity hardware.