Highlights d Cross-study meta-analysis of metagenomes covering 3,655 samples from two body sites d Meta-analysis uncovers staggering microbial gene diversity d 50% of all genes in a metagenomic sample are individualspecific or ''singletons'' d Individual's microbiomes can be fingerprinted via rare microbial strains
Polybrominated diphenyl ethers (PBDEs) are brominated flame retardants that act as endocrine disruptors, affecting thyroid hormone homeostasis. As a follow-up to a recent study showing high PBDE levels in household cats and linking PBDE levels with cat hyperthyroidism, we measured PBDEs, polychlorinated biphenyls (PCBs), and organochlorinated pesticides (OCPs) in serum samples from 26 California household cats (16 hyperthyroid, 10 controls) using liquid-liquid extraction and high-resolution gas chromatography/high-resolution mass spectrometry. In the present pilot study, we found that PBDE levels in California house cats were extremely high (ΣPBDEs median = 2,904 ng/g lipid; range, 631-22,537 ng/g lipid). This is approximately 50 times higher than levels in California residents (ΣPBDEs geomean = 62 ± 8.9 ng/g lipid, National Health and Nutrition Examination Survey), who have among the highest human levels in the world. Polybrominated diphenyl ethers congener patterns (BDE-99 major congener, BDE-209 significant) differed markedly from patterns found in California residents (BDE-47 major) or wildlife but resembled patterns found in house dust. Polychlorinated biphenyls and OCPs in cats were highly correlated, consistent with a shared dietary source or pathway of exposure, but did not correlate with PBDEs. This suggests a different source or pathway of exposure for PBDEs, which was most likely house dust. The authors found no evidence that linked levels of PBDEs, PCBs, or OCPs with hyperthyroidism. This may be because of the small sample size, competing or confounding risk factors, or complicated causal mechanisms.
We empirically show that the test error of deep networks can be estimated by simply training the same architecture on the same training set but with a different run of Stochastic Gradient Descent (SGD), and measuring the disagreement rate between the two networks on unlabeled test data. This builds onand is a stronger version of -the observation in Nakkiran & Bansal (2020), which requires the second run to be on an altogether fresh training set. We further theoretically show that this peculiar phenomenon arises from the well-calibrated nature of ensembles of SGD-trained models. This finding not only provides a simple empirical measure to directly predict the test error using unlabeled test data, but also establishes a new conceptual connection between generalization and calibration.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.