Pre-trained language representation models (PLMs) cannot well capture factual knowledge from text. In contrast, knowledge embedding (KE) methods can effectively represent the relational facts in knowledge graphs (KGs) with informative entity embeddings, but conventional KE models cannot take full advantage of the abundant textual information. In this paper, we propose a unified model for Knowledge Embedding and Pre-trained LanguagERepresentation (KEPLER), which can not only better integrate factual knowledge into PLMs but also produce effective text-enhanced KE with the strong PLMs. In KEPLER, we encode textual entity descriptions with a PLM as their embeddings, and then jointly optimize the KE and language modeling objectives. Experimental results show that KEPLER achieves state-of-the-art performances on various NLP tasks, and also works remarkably well as an inductive KE model on KG link prediction. Furthermore, for pre-training and evaluating KEPLER, we construct Wikidata5M1 , a large-scale KG dataset with aligned entity descriptions, and benchmark state-of-the-art KE methods on it. It shall serve as a new KE benchmark and facilitate the research on large KG, inductive KE, and KG with text. The source code can be obtained from https://github.com/THU-KEG/KEPLER.

Wearable sensing technology is an essential link to future personalized medicine. However, to obtain a complete picture of human health, it is necessary but challenging to track multiple analytes inside the body simultaneously. Here, we present a wearable plasmonic-electronic sensor with “universal” molecular recognition ability. Flexible plasmonic metasurface with surface-enhanced Raman scattering (SERS)–activity is introduced as the fundamental sensing component in a wearable sensor since we solved the technical challenge of maintaining the plasmonic activities of their brittle nanostructures under various deformations. Together with a flexible electronic sweat extraction system, our sensor can noninvasively extract and “fingerprint” analytes inside the body based on their unique SERS spectra. As a proof-of-concept example, we successfully monitored the variation of trace-amounts drugs inside the body and obtained an individual’s drug metabolic profile. Our sensor bridges the existing gap in wearable sensing technology by providing a universal, sensitive molecular tracking means to assess human health.

No abstract

Long non-coding RNAs (lncRNAs) are involved in various pathophysiologic processes and human diseases. However, their dynamics and corresponding functions in pulmonary fibrosis remain poorly understood. In this study, portions of lncRNAs adjacent or homologous to protein-coding genes were determined by searching the UCSC genome bioinformatics database. This was found to be potentially useful for exploring lncRNA functions in disease progression. Previous studies showed that competing endogenous RNA (ceRNA) hypothesis is another method to predict lncRNA function. However, little is known about the function of ceRNA in pulmonary fibrosis. In this study, we selected two differentially expressed lncRNAs MRAK088388 and MRAK081523 to explore their regulatory mechanisms. MRAK088388 and MRAK081523 were analysed as long-intergenic non-coding RNAs (lincRNAs), and identified as orthologues of mouse lncRNAs AK088388 and AK081523, respectively. qRT-PCR and in situ hybridization (ISH) showed that they were significantly up-regulated, and located in the cytoplasm of interstitial lung cells. We also showed that MRAK088388 and N4bp2 had the same miRNA response elements (MREs) for miR-200, miR-429, miR-29, and miR-30, whereas MRAK081523 and Plxna4 had the same MREs for miR-218, miR-141, miR-98, and let-7. Moreover, the expression levels of N4bp2 and Plxna4 significantly increased in fibrotic rats, and were highly correlated with those of MRAK088388 and MRAK081523, respectively. Among their shared miRNAs, miR-29b-3p and let-7i-5p decreased in the model group, and were negatively correlated with the expression of MRAK088388 and MRAK081523, respectively. MRAK088388 and MRAK081523 could regulate N4bp2 and Plxna4 expression by sponging miR-29b-3p and let-7i-5p, respectively, and possessed regulatory functions as ceRNAs. Thus, our study may provide insights into the functional interactions of lncRNA, miRNA and mRNA, and lead to new theories for the pathogenesis and treatment of pulmonary fibrosis.