Node-degree aware edge sampling mitigates inflated classification performance in biomedical random walk-based graph representation learning.

Motivation: Graph representation learning is a family of related approaches that learn low-dimensional vector representations of nodes and other graph elements called embeddings. Embeddings approximate characteristics of the graph and can be used for a variety of machine-learning tasks such as novel edge prediction. For many biomedical applications, partial knowledge exists about positive edges that represent relationships between pairs of entities, but little to no knowledge is available about negative edges that represent the explicit lack of a relationship between two nodes. For this reason, classification procedures are forced to assume that the vast majority of unlabeled edges are negative. Existing approaches to sampling negative edges for training and evaluating classifiers do so by uniformly sampling pairs of nodes. Results: We show here that this sampling strategy typically leads to sets of positive and negative examples with imbalanced node degree distributions. Using representative heterogeneous biomedical knowledge graph and random walk-based graph machine learning, we show that this strategy substantially impacts classification performance. If users of graph machine-learning models apply the models to prioritize examples that are drawn from approximately the same distribution as the positive examples are, then performance of models as estimated in the validation phase may be artificially inflated. We present a degree-aware node sampling approach that mitigates this effect and is simple to implement. Availability and implementation: Our code and data are publicly available at https://github.com/monarch-initiative/negativeExampleSelection.

Adipose retinol saturase is regulated by ß-adrenergic signaling and its deletion impairs lipolysis in adipocytes and acute cold tolerance in mice.

OBJECTIVE: Retinol saturase (RetSat) is an endoplasmic reticulum-localized oxidoreductase highly expressed in organs involved in lipid metabolism such as white (WAT) and brown adipose tissue (BAT). Cold exposure was shown to increase RETSAT protein in BAT but its relevance for non-shivering thermogenesis, a process with beneficial effects on metabolic health, is unknown. METHODS: We analyzed the regulation of RetSat expression in white and brown adipocytes and different murine adipose tissue depots upon ß-adrenergic stimulation and cold exposure. RetSat function during the differentiation and ß-adrenergic stimulation of brown adipocytes was dissected by loss-of-function experiments. Mice with BAT-specific deletion of RetSat were generated and exposed to cold. Gene expression in human WAT was analyzed and the effect of RetSat depletion on adipocyte lipolysis investigated. RESULTS: We show that cold exposure induces RetSat expression in both WAT and BAT of mice via ß-adrenergic signaling. In brown adipocytes, RetSat has minor effects on differentiation but is required for maximal thermogenic gene and protein expression upon ß-adrenergic stimulation and mitochondrial respiration. In mice, BAT-specific deletion of RetSat impaired acute but not long-term adaptation to cold exposure. RetSat expression in subcutaneous WAT of humans correlates with the expression of genes related to mitochondrial function. Mechanistically, we found that RetSat depletion impaired ß-agonist-induced lipolysis, a major regulator of thermogenic gene expression in adipocytes. CONCLUSIONS: Thus, RetSat expression is under ß-adrenergic control and determines thermogenic capacity of brown adipocytes and acute cold tolerance in mice. Modulating RetSat activity may allow for therapeutic interventions towards pathologies with inadequate metabolic activity.