Long-read and strand-specific sequencing technologies together facilitate the de novo assembly of high-quality haplotype-resolved human genomes without parent–child trio data. We present 64 assembled haplotypes from 32 diverse human genomes. These highly contiguous haplotype assemblies (average contig N50: 26 Mbp) integrate all forms of genetic variation even across complex loci. We identify 107,590 structural variants (SVs), of which 68% are not discovered by short-read sequencing, and 278 SV hotspots (spanning megabases of gene-rich sequence). We characterize 130 of the most active mobile element source elements and find that 63% of all SVs arise by homology-mediated mechanisms. This resource enables reliable graph-based genotyping from short reads of up to 50,340 SVs, resulting in the identification of 1,526 expression quantitative trait loci as well as SV candidates for adaptive selection within the human population.

No abstract

For millions of years, plants evolve plenty of structurally diverse secondary metabolites (SM) to support their sessile lifestyles through continuous biochemical pathway innovation. While new genes commonly drive the evolution of plant SM pathway, how a full biosynthetic pathway evolves remains poorly understood. The evolution of pathway involves recruiting new genes along the reaction cascade forwardly, backwardly, or in a patchwork manner. With three chromosome-scale Papaver genome assemblies, we here reveal whole-genome duplications (WGDs) apparently accelerate chromosomal rearrangements with a nonrandom distribution towards SM optimization. A burst of structural variants involving fusions, translocations and duplications within 7.7 million years have assembled nine genes into the benzylisoquinoline alkaloids gene cluster, following a punctuated patchwork model. Biosynthetic gene copies and their total expression matter to morphinan production. Our results demonstrate how new genes have been recruited from a WGD-induced repertoire of unregulated enzymes with promiscuous reactivities to innovate efficient metabolic pathways with spatiotemporal constraint.

Deep learning models are vulnerable to adversarial examples crafted by applying human-imperceptible perturbations on benign inputs. However, under the blackbox setting, most existing adversaries often have a low transferability to attack other defense models. In this work, from the perspective of regarding the adversarial example generation as an optimization process, we propose two new methods to improve the transferability of adversarial examples, namely Nesterov Iterative Fast Gradient Sign Method (NI-FGSM) and Scale-Invariant attack Method (SIM). NI-FGSM aims to adapt Nesterov accelerated gradient into the iterative attacks so as to effectively look ahead and avoid the "missing" of the global maximum. While SIM is based on our discovery on the scale-invariant property of deep learning models, for which we leverage to optimize the adversarial perturbations over the scale copies of the input images so as to avoid "overfitting" on the whitebox model being attacked and generate more transferable adversarial examples. NI-FGSM and SIM can be naturally integrated to build a robust gradient-based attack to generate more transferable adversarial examples against the defense models. Empirical results on ImageNet dataset and NIPS 2017 adversarial competition demonstrate that our attack methods exhibit higher transferability and achieve higher attack success rates than state-of-the-art gradient-based attacks.