On-target activities and genome-wide specificities of Cpf1 nucleases in human cells. The activities and genome-wide specificities of CRISPR-Cas Cpf1 nucleases1 are not well defined. We show that two Cpf1 nucleases from Acidaminococcus sp. BV3L6 and Lachnospiraceae bacterium ND2006 (AsCpf1 and LbCpf1, respectively) have on-target efficiencies in human cells comparable with those of the widely used Streptococcus pyogenes Cas9 (SpCas9)2,3,4,5. We also report that four to six bases at the 3′ end of the short CRISPR RNA (crRNA) used to program Cpf1 nucleases are insensitive to single base mismatches, but that many of the other bases in this region of the crRNA are highly sensitive to single or double substitutions. Using GUIDE-seq and targeted deep sequencing analyses performed with both Cpf1 nucleases, we were unable to detect off-target cleavage for more than half of 20 different crRNAs. Our results suggest that AsCpf1 and LbCpf1 are highly specific in human cells.

Significance: Understanding dendritic cell (DC) migration during an immune response is fundamental to defining the rules that govern T cell-mediated immunity. We recently described mice deficient in the pattern recognition receptor NLRP10 (NLR family, pyrin domain containing 10) with a severe DC migration defect. Using whole-exome sequencing, we discovered that this defect was due to mutation of the guanine nucleotide exchange factor Dock8 (dedicator of cytokinesis 8). DOCK8 regulates cytoskeleton dynamics in leukocytes, and loss-of-function mutations cause an immunodeficiency syndrome. Mutations in other Dock genes have been reported in mice lacking innate immune pathways, and we now report two more lines with Dock8 mutations resulting in impaired DC migration. These results clarify the role of NLRP10 in DCs and confirm the essential function of DOCK8 in the immune system.

Although high-dimensional clinical data (HDCD) are increasingly available in biobank-scale datasets, their use for genetic discovery remains challenging. Here we introduce an unsupervised deep learning model, Representation Learning for Genetic Discovery on Low-Dimensional Embeddings (REGLE), for discovering associations between genetic variants and HDCD. REGLE leverages variational autoencoders to compute nonlinear disentangled embeddings of HDCD, which become the inputs to genome-wide association studies (GWAS). REGLE can uncover features not captured by existing expert-defined features and enables the creation of accurate disease-specific polygenic risk scores (PRSs) in datasets with very few labeled data. We apply REGLE to perform GWAS on respiratory and circulatory HDCD—spirograms measuring lung function and photoplethysmograms measuring blood volume changes. REGLE replicates known loci while identifying others not previously detected. REGLE are predictive of overall survival, and PRSs constructed from REGLE loci improve disease prediction across multiple biobanks. Overall, REGLE contain clinically relevant information beyond that captured by existing expert-defined features, leading to improved genetic discovery and disease prediction.