The field of single-cell analysis and knowledge graph embeddings is rapidly advancing, with a focus on developing innovative methods for modeling cellular responses to various treatments and predicting gene-disease associations. Recent research has highlighted the potential of neural optimal transport and conditional models in predicting single-cell perturbation responses, as well as the importance of incorporating semantic richness and disease-specific ontologies in knowledge graph embeddings. Notably, the use of contrastive learning techniques and subset-contrastive approaches has shown promise in enhancing the analysis of omics data and improving the accuracy of gene regulatory network inference. Overall, the field is moving towards the development of more generalizable and scalable methods that can accommodate the complexities of single-cell data and knowledge graphs. Noteworthy papers include:

• A paper proposing the Conditional Monge Gap, which achieves state-of-the-art results in predicting single-cell perturbation responses conditional to one or multiple drugs.
• A paper introducing a novel framework for comparing the performance of link prediction versus node-pair classification tasks in gene-disease association prediction, showing that link prediction methods better explore the semantic richness encoded in knowledge graphs.