Understanding complex interactions in biomedical networks is crucial for advancements in biomedicine, but traditional link prediction (LP) methods are limited in capturing this complexity. We present BioPathNet, a graph neural network framework based on the neural Bellman-Ford network (NBFNet), addressing limitations of traditional representation-based learning methods through path-based reasoning for LP in biomedical knowledge graphs. Unlike node-embedding frameworks, BioPathNet learns representations between node pairs by considering all relations along paths, enhancing prediction accuracy and interpretability, and allowing visualization of influential paths and biological validation. BioPathNet leverages a background regulatory graph for enhanced message passing and uses stringent negative sampling to improve precision and scalability. BioPathNet outperforms or matches existing methods across diverse tasks including gene function annotation, drug-disease indication, synthetic lethality and lncRNA-target interaction prediction. Our study identifies promising additional drug indications for diseases such as acute lymphoblastic leukaemia and Alzheimer's disease, validated by medical experts and clinical trials. In addition, we prioritize putative synthetic lethal gene pairs and regulatory lncRNA-target interactions. BioPathNet's interpretability will enable researchers to trace prediction paths and gain molecular insights.

France and Spain have been fighting against bovine tuberculosis (bTB) for years, even though new outbreaks continue to appear on both sides of the border, generating misconceptions about the disease and social distrust in the eradication programme and competent authorities. The perceived disease risk and the commitment of all interested parties are key factors for the successful implementation of control programmes, as they might influence the acceptability of recommended measures. Effective communication can contribute to increasing knowledge, trust and stakeholders' engagement, thus ensuring the acceptability of the eradication programme. This study was conducted in Catalonia (Spain) and Pyrenees-Atlantiques (France) in the frame of the INNOTUB project (https://innotub.eu/) to characterise the communication on bTB in the trans-Pyrenees region and provide recommendations to improve it. The communication on bTB was characterised by analysing 153 (Spain) and 66 (France) online freely available texts, published between 2018 and 2020, through Content Analysis and Critical Metaphor Analysis. Moreover, six farmers and four veterinarians were in-depth interviewed in each area to gather information about the communication on bTB. Interviews were made in original languages and analysed using a qualitative thematic approach. A pilot participatory intervention inspired by the Systematic Tool for Behavioural Assumption Validation and Exploration (STAVE) method was used to develop a list of proposals to improve communication and to promote the creation of territorial networks/committees on bTB prevention and control. It included three focus groups with farmers and veterinarians, a meeting with representatives of the regional veterinary services, and a final deliberative workshop. Results highlight the existence of a controversial debate on bTB and a heterogeneous understanding between stakeholders. Institutional and scientific communication mainly focus on bTB detection and control while other aspects are left in the background. On the contrary, farmers extend their communication to a greater variety of topics. The metaphorical framing strongly differed among actors, while veterinary services and researchers "fight" against bTB and "progress" toward the eradication, farmers place themselves in a framework of "sacrifice" and, particularly in Spain, they play a passive role. The proposals developed by the participants to improve the current communication on bTB included: (i) create participatory meeting spaces to share opinions and information; (ii) improve data accessibility (on epidemiological situations); (iii) develop clearer and written protocols and informative visual material; (iv) redesign the training courses (v) increase the stakeholders' participation in the design of protocols.

We present RBPNet, a novel deep learning method, which predicts CLIP-seq crosslink count distribution from RNA sequence at single-nucleotide resolution. By training on up to a million regions, RBPNet achieves high generalization on eCLIP, iCLIP and miCLIP assays, outperforming state-of-the-art classifiers. RBPNet performs bias correction by modeling the raw signal as a mixture of the protein-specific and background signal. Through model interrogation via Integrated Gradients, RBPNet identifies predictive sub-sequences that correspond to known and novel binding motifs and enables variant-impact scoring via in silico mutagenesis. Together, RBPNet improves imputation of protein-RNA interactions, as well as mechanistic interpretation of predictions.

The iCLIP and eCLIP techniques facilitate the detection of protein-RNA interaction sites at high resolution, based on diagnostic events at crosslink sites. However, previous methods do not explicitly model the specifics of iCLIP and eCLIP truncation patterns and possible biases. We developed PureCLIP ( https://github.com/skrakau/PureCLIP ), a hidden Markov model based approach, which simultaneously performs peak-calling and individual crosslink site detection. It explicitly incorporates a non-specific background signal and, for the first time, non-specific sequence biases. On both simulated and real data, PureCLIP is more accurate in calling crosslink sites than other state-of-the-art methods and has a higher agreement across replicates.