High-quality reference genomes for the platypus and echidna — two egg-laying mammals, or monotremes, of Australasia — are presented in Nature this week, providing insights into mammalian evolution. An international team led by scientists from the University of Copenhagen generated the platypus genome using a combination of single-molecule sequencing technology and physical mapping methods, assigning most of the sequences to a chromosome-scale assembly and markedly improving the genome continuity and gene annotation. A less-continuous assembly was also produced for the short-beaked echidna. In analyzing the genomes, the researchers were able to detect the ancestral and lineage-specific genomic changes shaping both monotreme and mammalian evolution. "The new genomes of both species will enable further insights into therian innovations and the biology and evolution of these extraordinary egg-laying mammals," the authors write.
In performing single-cell RNA sequencing of tumor cells from patients with gastric adenocarcinoma, a team of MD Anderson Cancer Center scientists have identified key contributors to intratumoral heterogeneity in cancer. As reported in Nature Medicine, the investigators sequenced peritoneal carcinomatosis (PC) cells from 15 patients to construct a map of over 45,000 PC cells, profile the transcriptome states of tumor cell populations, examine the intratumoral heterogeneity of malignant PC cells, and identify significant correlates with patient survival. They determine that diversity in tumor cell lineage/state compositions is a key contributor to intratumoral heterogeneity, as well as discover a 12-gene prognostic signature that may hold potential for stratifying patients for treatment.
A new method for improved metagenome binning and assembly is described in Nature Biotechnology this week. Metagenomic binning, the process of grouping metagenomic sequences by their organism of origin, is a key starting point for metagenomic studies, but the reconstruction of microbial species from metagenomics data remains challenging. To address this, researchers from the University of Copenhagen and collaborators developed a computational tool — called variational autoencoders for metagenomic binning, or VAMB — that uses deep variational autoencoders to encode sequence co-abundance and k-mer distribution information before clustering. The team shows that VAMB is able to integrate these two distinct data types without any previous knowledge of the datasets and can outperform existing state-of-the-art binners.