Nature Biomedical Engineering最新文献

Clinical and biological information in large datasets of gene expression across cancers could be tapped with unsupervised deep learning. However, difficulties associated with biological interpretability and methodological robustness have made this impractical. Here we describe an unsupervised deep-learning framework for the generation of low-dimensional latent spaces for gene-expression data from 50,211 transcriptomes across 18 human cancers. The framework, which we named DeepProfile, outperformed dimensionality-reduction methods with respect to biological interpretability and allowed us to unveil that genes that are universally important in defining latent spaces across cancer types control immune cell activation, whereas cancer-type-specific genes and pathways define molecular disease subtypes. By linking latent variables in DeepProfile to secondary characteristics of tumours, we discovered that mutation burden is closely associated with the expression of cell-cycle-related genes, and that the activity of biological pathways for DNA-mismatch repair and MHC class II antigen presentation are consistently associated with patient survival. We also found that tumour-associated macrophages are a source of survival-correlated MHC class II transcripts. Unsupervised learning can facilitate the discovery of biological insight from gene-expression data.

Circular RNA (circRNA) is a candidate for next-generation messenger RNA therapeutics owing to its remarkable stability. Here we describe trans-splicing-based methods for the synthesis of circRNAs over 8,000 nucleotides. The methods are independent of bacterial sequences, outperform the permuted intron–exon method and allow for the incorporation of RNA modifications. The resulting unmodified circRNAs, which incorporate sequences from human 28S ribosomal RNA, display low immunogenicity and are translated more efficiently than permuted intron–exon-derived circRNAs. Additionally, by using viral internal ribosomal entry sites for rolling circle translation, we show that ribosomes can efficiently read through highly structured internal ribosomal entry sites, enhancing the efficiency of rolling circle translation by over 7,000-fold with respect to previous constructs. The efficient and reliable production of circRNA may facilitate its therapeutic use.

In patients with pancreatic ductal adenocarcinoma (PDAC), intratumoural and intertumoural heterogeneity increases chemoresistance and mortality rates. However, such morphological and phenotypic diversities are not typically captured by organoid models of PDAC. Here we show that branched organoids embedded in collagen gels can recapitulate the phenotypic landscape seen in murine and human PDAC, that the pronounced molecular and morphological intratumoural and intertumoural heterogeneity of organoids is governed by defined transcriptional programmes (notably, epithelial-to-mesenchymal plasticity), and that different organoid phenotypes represent distinct tumour-cell states with unique biological features in vivo. We also show that phenotype-specific therapeutic vulnerabilities and modes of treatment-induced phenotype reprogramming can be captured in phenotypic heterogeneity maps. Our methodology and analyses of tumour-cell heterogeneity in PDAC may guide the development of phenotype-targeted treatment strategies.

To infer intent, brain–computer interfaces must extract features that accurately estimate neural activity. However, the degradation of signal quality over time hinders the use of feature-engineering techniques to recover functional information. By using neural data recorded from electrode arrays implanted in the cortices of three human participants, here we show that a convolutional neural network can be used to map electrical signals to neural features by jointly optimizing feature extraction and decoding under the constraint that all the electrodes must use the same neural-network parameters. In all three participants, the neural network led to offline and online performance improvements in a cursor-control task across all metrics, outperforming the rate of threshold crossings and wavelet decomposition of the broadband neural data (among other feature-extraction techniques). We also show that the trained neural network can be used without modification for new datasets, brain areas and participants.