In Nature Genetics this week, an international research team presents a large genome-wide association study of Alzheimer's disease, identifying new loci and pathways that influence disease risk. The study includes 71,880 Alzheimer's cases and more than 380,000 controls, and uncovers 29 risk loci implicating 215 potential causative genes. The researchers also demonstrate strong genetic correlations with multiple health-related outcomes and find a protective effect of cognitive ability on Alzheimer's disease risk. "The results of this study could ... serve as a valuable resource for selection of promising genes for functional follow-up experiments and identifying targets for drug development and stratification approaches," they write. GenomeWeb has more on this, here.
And in Nature Medicine, researchers from facial analysis firm FDNA and academic collaborators report data showing that an artificial intelligence approach can identify rare genetic syndromes with high accuracy after being trained on facial images of patients. Using more than 17,000 facial images representing over 200 distinct genetic syndromes, the phenotyping framework — dubbed DeepGestalt — was used to analyze 502 different images of patients with clinical diagnoses. The AI was able to offer the correct syndrome within its top 10 suggestions of potential disorders 91 percent of the time. The researchers say their approach could simplify the identification of undiagnosed patients, but stress the need for monitoring strategies to prevent abuse given the ease with which facial images can be accessed.
Meanwhile, in Nature Microbiology, a group of US and Chinese investigators describes a new CRISPR interference-based approach that can be used for genetic analysis of diverse bacteria. Called Mobile-CRISPRi, it comprises a suite of CRISPR interference systems combining modularity, stable genomic integration, and ease of transfer to diverse bacteria by conjugation, the authors write. They demonstrate Mobile-CRISPRi on gammaproteobacteria and Bacillales Firmicutes at the individual gene and library scale, and predict their technique will be "a transformative technology for non-model bacteria lacking genetic tools and will facilitate cross-species genetic analysis."