Are we fitting data or noise? Analysing the predictive power of commonly used datasets in drug-, materials-, and molecular-discovery.

Data-driven techniques for establishing quantitative structure property relations are a pillar of modern materials and molecular discovery. Fuelled by the recent progress in deep learning methodology and the abundance of new algorithms, it is tempting to chase benchmarks and incrementally build ever more capable machine learning (ML) models. While model evaluation has made significant progress, the intrinsic limitations arising from the underlying experimental data are often overlooked.

Proteomics of mouse brain endothelium uncovers dysregulation of vesicular transport pathways during aging.

Age-related decline in brain endothelial cell (BEC) function contributes critically to neurological disease. Comprehensive atlases of the BEC transcriptome have become available, but results from proteomic profiling are lacking. To gain insights into endothelial pathways affected by aging, we developed a magnetic-activated cell sorting-based mouse BEC enrichment protocol compatible with proteomics and resolved the profiles of protein abundance changes during aging.

Reproducible and scalable differentiation of highly pure cortical neurons from human induced pluripotent stem cells.

Human-induced-pluripotent-stem-cell (hiPSC)-derived neurons are valuable for investigating brain physiology and disease. Here, we present a protocol to differentiate hiPSCs into cortical neurons with high yield and purity. We describe neural induction via dual-SMAD inhibition, followed by spot-based differentiation to provide high quantities of neural precursors.