Intellectual disability (ID) is a pathological condition characterized by limited intellectual functioning and adaptive behaviors. It affects 1–3% of the worldwide population, and no pharmacological therapies are currently available. More than 1000 genes have been found mutated in ID patients pointing out that, despite the common phenotype, the genetic bases are highly heterogeneous and apparently unrelated. Bibliomic analysis reveals that ID genes converge onto a few biological modules, including cytoskeleton dynamics, whose regulation depends on Rho GTPases transduction. Genetic variants exert their effects at different levels in a hierarchical arrangement, starting from the molecular level and moving toward higher levels of organization, i.e., cell compartment and functions, circuits, cognition, and behavior. Thus, cytoskeleton alterations that have an impact on cell processes such as neuronal migration, neuritogenesis, and synaptic plasticity rebound on the overall establishment of an effective network and consequently on the cognitive phenotype. Systems biology (SB) approaches are more focused on the overall interconnected network rather than on individual genes, thus encouraging the design of therapies that aim to correct common dysregulated biological processes. This review summarizes current knowledge about cytoskeleton control in neurons and its relevance for the ID pathogenesis, exploiting in silico modeling and translating the implications of those findings into biomedical research.

In recent years, the debate in the field of applications of Deep Learning to Virtual Screening has focused on the use of neural embeddings with respect to classical descriptors in order to encode both structural and physical properties of ligands and/or targets. The attention on embeddings with the increasing use of Graph Neural Networks aimed at overcoming molecular fingerprints that are short range embeddings for atomic neighborhoods. Here, we present EMBER, a novel molecular embedding made by seven molecular fingerprints arranged as different “spectra” to describe the same molecule, and we prove its effectiveness by using deep convolutional architecture that assesses ligands’ bioactivity on a data set containing twenty protein kinases with similar binding sites to CDK1. The data set itself is presented, and the architecture is explained in detail along with its training procedure. We report experimental results and an explainability analysis to assess the contribution of each fingerprint to different targets.

Background A Virtual Screening algorithm has to adapt to the different stages of this process. Early screening needs to ensure that all bioactive compounds are ranked in the first positions despite of the number of false positives, while a second screening round is aimed at increasing the prediction accuracy. Results A novel CNN architecture is presented to this aim, which predicts bioactivity of candidate compounds on CDK1 using a combination of molecular fingerprints as their vector representation, and has been trained suitably to achieve good results as regards both enrichment factor and accuracy in different screening modes (98.55% accuracy in active-only selection, and 98.88% in high precision discrimination). Conclusion The proposed architecture outperforms state-of-the-art ML approaches, and some interesting insights on molecular fingerprints are devised.

No abstract

No abstract