Computational biology and bioinformatics

A time-resolved high-resolution map of human cardiac remodelling after myocardial infarction, integrating single-cell transcriptomic, chromatin accessibility and spatial transcriptomic data, provides a valuable resource for the field.

Copy number variations inferred from spatial transcriptomics data in benign and malignant tissue reveal clonal architecture at the organ-wide level.

To understand the contribution of variants to transcript expression regulation, long-read transcriptome data are generated from the GTEx resource, and a new software package to perform allele-specific analysis is developed.

Global ocean microbiome survey reveals the bacterial family ‘Candidatus Eudoremicrobiaceae’, which includes some of the most biosynthetically diverse microorganisms in the ocean environment.

Copy number signatures characterize different types of chromosomal instability and predict drug response.

A new framework enables a pan-cancer reference set of copy number signatures derived from allele-specific profiles from different experimental assays.

A long-term study of 385 human donors reports that driver gene mutations and age determine the lifelong dynamics of clonal haematopoiesis

A computational system termed MetaWIBELE (workflow to identify novel bioactive elements in the microbiome) is used to identify microbial gene products that are potentially bioactive and have a functional role in the pathogenesis of inflammatory bowel disease.

Single-nucleus Hi-C of embryos, polymer simulations and single-molecule imaging collectively reveal that MCM complexes influence genome folding and gene expression by impeding DNA loop extrusion.

A single-cell Poisson model is used to analyse eQTLs in memory T cells across continuous, dynamic cell states, revealing that the cell context is critical to understanding variation in eQTLs and their association with disease.

A mathematical framework to estimate the fitness of cancer driver mutations by integrating mutational bias, oncogenicity and immunogenicity finds fundamental trade-offs in cancer evolution.

The Human Pangenome Reference Consortium aims to offer the highest quality and most complete human pangenome reference that provides diverse genomic representation across human populations.

A study in which gut microbiomes of 8,208 individuals from 2,756 families were characterized and correlated to 241 host and environmental factors defines microbiome patterns shared across diverse diseases and shows that the microbiome is shaped largely by environment and cohabitation.

A large-scale single-cell transcriptomic atlas of the non-human primate Macaca fascicularis encompasses over 1 million cells from 45 adult tissues.

Matched Annotation from NCBI and EMBL-EBI (MANE) delivers joint transcript sets from Ensembl/GENCODE and RefSeq for standardizing variant reporting in clinical genomics and research.

A design pipeline is presented whereby binding proteins can be designed de novo without the need for prior information on binding hotspots or fragments from structures of complexes with binding partners.

A framework for studying and engineering gene regulatory DNA sequences, based on deep neural sequence-to-expression models trained on large-scale libraries of random DNA, provides insight into the evolution, evolvability and fitness landscapes of regulatory DNA.

Analyses of circulating cell-free RNA (cfRNA) in blood samples from pregnant mothers identify changes in gene expression that could be used in liquid biopsy tests to identify and monitor individuals who are at risk of preeclampsia.

Modelling by SCUBA of the backbone-centred energy surface extends the diversity of designable proteins.

Serratus, an open-source cloud-computing infrastructure, can be used to screen millions of nucleic acid sequencing libraries at the petabase scale, and has enabled many new RNA viruses to be identified efficiently.

A chromatin accessibility atlas of 240,919 cells in the adult and developing Drosophila brain reveals 95,000 enhancers, which are integrated in cell-type specific enhancer gene regulatory networks and decoded into combinations of functional transcription factor binding sites using deep learning.

Bayesian analysis of datasets comprising genomes from multiple mammalian species can efficiently and precisely decipher their evolutionary timeline.

A survey of species-level genes from 13,174 publicly available metagenomes shows that most species-level genes are specific to a single habitat, encode a small number of protein families and are under low positive (adaptive) pressure.

A genome-wide association meta-analysis study of blood lipid levels in roughly 1.6 million individuals demonstrates the gain of power attained when diverse ancestries are included to improve fine-mapping and polygenic score generation, with gains in locus discovery related to sample size.

Crystal structures of the σ₂ receptor are determined and used to perform a docking screen of nearly 500 million molecules, identifying σ₂-selective ligands and providing insight into the role of σ₂ in neuropathic pain.

Integration of pre-treatment tumour features in predictive models using machine learning could inform on response to therapy.

The trRosetta neural network was used to iteratively optimise model proteins from random 100-amino-acid sequences, resulting in ‘hallucinated’ proteins, which when expressed in bacteria closely resembled the model structures.

Protein immunofluorescence imaging and affinity purification–mass spectrometry are combined to create a unified map of human cell architecture across scales, which the authors call the multi-scale integrated cell (MuSIC).

The presence of porcine deltacoronavirus has been detected in three children from Haiti that could have originated from zoonotic spillover.

A new technique called immunoGAM, which combines genome architecture mapping (GAM) with immunoselection, enabled the discovery of specialized chromatin conformations linked to gene expression in specific cell populations from mouse brain tissues.

Serial femtosecond crystallography (SFX) has provided significant understanding of time-resolved processes of various systems in biology, for example, rhodopsin, which underlies our vision. The approach involves femtosecond-length X-ray pulses directed at protein crystals and has been used to study various photoactive proteins. However, the function of proteins such as rhodopsin requires trans–cis isomerization of a chromophore, which involves crossing of a conical intersection—a funnel separating potential energy surfaces—at timescales faster than what can be achieved experimentally. Here, Ourmazd and colleagues report a machine learning analysis of SFX data of photoactive yellow protein, which resolves the protein passing through a conical intersection, providing information about the potential energy surfaces involved and achieving time resolution of less than 10 fs. This approach offers an opportunity to understand some of the fastest processes in biology by extracting even more information from SFX datasets.

A new computational method, EVE, classifies human genetic variants in disease genes using deep generative models trained solely on evolutionary sequences.

Modelling highlights international travel as the main driver of the introduction of SARS-CoV-2 to Europe and the USA, and suggests that introductions and local transmission may have begun in January 2020.

Focused ion beam scanning electron microscopy (FIB-SEM) combined with deep-learning-based segmentation is used to produce three-dimensional reconstructions of complete cells and tissues, in which up to 35 different organelle classes are annotated.

Analysis of genomic networks from 430 modern individuals across 21 Pacific island populations reveals the human settlement history of Polynesia.

A robust, cost-effective technique based on whole-exome sequencing data can be used to characterize immune infiltrates, relate the extent of these infiltrates to somatic changes in tumours, and enables prediction of tumour responses to immune checkpoint inhibition therapy.

Cells from embryonic, fetal, paediatric and adult human intestinal tissue are analysed at different locations along the intestinal tract to construct a single-cell atlas of the developing and adult human intestinal tract, encompassing all cell lineages.

Data-driven modelling including numbers of cases and population movements is used to simulate the COVID-19 pandemic in the United States in 2020, providing insights into the transmission of the disease.

Adult human tissues from diverse sites around the body are used to reconstruct cellular phylogenies from early development, using somatic mutations as an internal barcode.

This review describes the state of spatial transcriptomics technologies and analysis tools that are being used to generate biological insights in diverse areas of biology.

AlphaFold is used to predict the structures of almost all of the proteins in the human proteome—the availability of high-confidence predicted structures could enable new avenues of investigation from a structural perspective.

High-performance hydrodynamic simulations show that the skeletal structure of the deep-sea sponge Euplectella aspergillum reduces the hydrodynamic stresses on it, while possibly being beneficial for feeding and reproduction.

AlphaFold predicts protein structures with an accuracy competitive with experimental structures in the majority of cases using a novel deep learning architecture.

The use of new datastreams and local knowledge to shed light on social aspects of disease transmission will allow more accurate modelling and prediction of epidemics.

A proximity-dependent biotinylation technique defines the location of more than 4,000 proteins in a human cell, and almost 36,000 proximal interactions between proteins, including those at the interface of the mitochondria and ER.