Search

Alzheimer’s disease (AD)—a complex disease showing multiple pathomechanisticalterations—is triggered by nonlinear dynamic interactions of genetic/epigeneticand environmental risk factors, which, ultimately, converge into a biologicallyheterogeneous disease. To tackle the burden of AD during early preclinicalstages, accessible blood-based biomarkers are currently being developed.Specifically, next-generation clinical trials are expected to integrate positiveand negative predictive blood-based biomarkers into study designs to evaluate,at the individual level, target druggability and potential drug resistancemechanisms. In this scenario, systems biology holds promise to acceleratevalidation and qualification for clinical trial contexts of use—includingproof-of-mechanism, patient selection, assessment of treatment efficacy andsafety rates, and prognostic evaluation. Albeit in their infancy, systemsbiology-based approaches are poised to identify relevant AD “signatures” throughmultifactorial and interindividual variability, allowing us to decipher diseasepathophysiology and etiology. Hopefully, innovative biomarker-drug codevelopmentstrategies will be the road ahead towards effective disease-modifyingdrugs.

Alzheimer’s disease (AD) is highly heritable and recent studies have identified over 20 disease-associated genomic loci. Yet these only explain a small proportion of the genetic variance, indicating that undiscovered loci remain. Here, we performed a large genome-wide association study of clinically diagnosed AD and AD-by-proxy (71,880 cases, 383,378 controls). AD-by-proxy, based on parental diagnoses, showed strong genetic correlation with AD (rg= 0.81). Meta-analysis identified 29 risk loci, implicating 215 potential causative genes. Associated genes are strongly expressed in immune-related tissues and cell types (spleen, liver, and microglia). Gene-set analyses indicate biological mechanisms involved in lipid-related processes and degradation of amyloid precursor proteins. We show strong genetic correlations with multiple health-related outcomes, and Mendelian randomization results suggest a protective effect of cognitive ability on AD risk. These results are a step forward in identifying the genetic factors that contribute to AD risk and add novel insights into the neurobiology of AD.

High-throughput screening (HTS), as one of the key techniques in drug discovery, is frequently used to identify promising drug candidates in a largely automated and cost-effective way. One of the necessary conditions for successful HTS campaigns is a large and diverse compound library, enabling hundreds of thousands of activity measurements per project. Such collections of data hold great promise for computational and experimental drug discovery efforts, especially when leveraged in combination with modern deep learning techniques, and can potentially lead to improved drug activity predictions and cheaper and more effective experimental design. However, existing collections of machine-learning-ready public datasets do not exploit the multiple data modalities present in real-world HTS projects. Thus, the largest fraction of experimental measurements, corresponding to hundreds of thousands of “noisy” activity values from primary screening, are effectively ignored in the majority of machine learning models of HTS data. To address these limitations, we introduce Multifidelity PubChem BioAssay (MF-PCBA), a curated collection of 60 datasets that includes two data modalities for each dataset, corresponding to primary and confirmatory screening, an aspect that we call multifidelity. Multifidelity data accurately reflect real-world HTS conventions and present a new, challenging task for machine learning: the integration of low- and high-fidelity measurements through molecular representation learning, taking into account the orders-of-magnitude difference in size between the primary and confirmatory screens. Here we detail the steps taken to assemble MF-PCBA in terms of data acquisition from PubChem and the filtering steps required to curate the raw data. We also provide an evaluation of a recent deep-learning-based method for multifidelity integration across the introduced datasets, demonstrating the benefit of leveraging all HTS modalities, and a discussion in terms of the roughness of the molecular activity landscape. In total, MF-PCBA contains over 16.6 million unique molecule–protein interactions. The datasets can be easily assembled by using the source code available at https://github.com/davidbuterez/mf-pcba.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.