A fast, scalable and versatile tool for analysis of single-cell omics data

Nature Methods, Journal Year: 2024, Volume and Issue: 21(2), P. 217 - 227

Published: Jan. 8, 2024

Abstract Single-cell omics technologies have revolutionized the study of gene regulation in complex tissues. A major computational challenge analyzing these datasets is to project large-scale and high-dimensional data into low-dimensional space while retaining relative relationships between cells. This low dimension embedding necessary decompose cellular heterogeneity reconstruct cell-type-specific regulatory programs. Traditional dimensionality reduction techniques, however, face challenges efficiency comprehensively addressing diversity across varied molecular modalities. Here we introduce a nonlinear algorithm, embodied Python package SnapATAC2, which not only achieves more precise capture single-cell heterogeneities but also ensures efficient runtime memory usage, scaling linearly with number Our algorithm demonstrates exceptional performance, scalability versatility diverse datasets, including assay for transposase-accessible chromatin using sequencing, RNA Hi-C multi-omics underscoring its utility advancing analysis.

Language: Английский

---

Joint dimension reduction and clustering analysis of single-cell RNA-seq and spatial transcriptomics data

Nucleic Acids Research, Journal Year: 2022, Volume and Issue: 50(12), P. e72 - e72

Published: March 22, 2022

Dimension reduction and (spatial) clustering is usually performed sequentially; however, the low-dimensional embeddings estimated in dimension-reduction step may not be relevant to class labels inferred step. We therefore developed a computation method, Dimension-Reduction Spatial-Clustering (DR-SC), that can simultaneously perform dimension within unified framework. Joint analysis by DR-SC produces accurate results ensures effective extraction of biologically informative features. applicable spatial transcriptomics characterizes organization tissue segregating it into multiple structures. Here, relies on latent hidden Markov random field model encourage smoothness detected cluster boundaries. Underlying an efficient expectation-maximization algorithm based iterative conditional mode. As such, scalable large sample sizes optimize parameter data-driven manner. With comprehensive simulations real data applications, we show outperforms existing methods: extracts more features than conventional methods, improves performance, offers improved trajectory inference visualization for downstream analyses.

Language: Английский

---

Collective variable discovery in the age of machine learning: reality, hype and everything in between

RSC Advances, Journal Year: 2022, Volume and Issue: 12(38), P. 25010 - 25024

Published: Jan. 1, 2022

Understanding the kinetics and thermodynamics profile of biomolecules is necessary to understand their functional roles which has a major impact in mechanism driven drug discovery. Molecular dynamics simulation been routinely used conformational molecular recognition biomolecules. Statistical analysis high-dimensional spatiotemporal data generated from requires identification few low-dimensional variables can describe essential system without significant loss information. In physical chemistry, these are often called collective variables. Collective generate reduced representations free energy surfaces calculate transition probabilities between different metastable basins. However choice not trivial for complex systems. range geometric criteria such as distances dihedral angles abstract ones weighted linear combinations multiple The advent machine learning algorithms led increasing use represent biomolecular dynamics. this review, I will highlight several nuances commonly ranging ones. Further, put forward some cases where based were simple systems principle could have described by Finally, my thoughts on artificial general intelligence how it be discover predict simulations.

Language: Английский

---

RNAincoder: a deep learning-based encoder for RNA and RNA-associated interaction

Nucleic Acids Research, Journal Year: 2023, Volume and Issue: 51(W1), P. W509 - W519

Published: May 11, 2023

Ribonucleic acids (RNAs) involve in various physiological/pathological processes by interacting with proteins, compounds, and other RNAs. A variety of powerful computational methods have been developed to predict such valuable interactions. However, all these rely heavily on the 'digitalization' (also known as 'encoding') RNA-associated pairs into a computer-recognizable descriptor. In words, it is urgently needed tool that can not only represent each partner but also integrate both partners interaction. Herein, RNAincoder (deep learning-based encoder for interactions) was therefore proposed (a) provide comprehensive collection RNA encoding features, (b) realize representation any interaction based well-established deep embedding strategy (c) enable large-scale scanning possible feature combinations identify one optimal performance prediction. The effectiveness extensively validated case studies benchmark datasets. All all, distinguished its capability providing more accurate interactions, which makes an indispensable complement available tools. be accessed at https://idrblab.org/rnaincoder/.

Language: Английский

---

Origin and segregation of the human germline

Life Science Alliance, Journal Year: 2023, Volume and Issue: 6(8), P. e202201706 - e202201706

Published: May 22, 2023

Human germline-soma segregation occurs during weeks 2-3 in gastrulating embryos. Although direct studies are hindered, here, we investigate the dynamics of human primordial germ cell (PGCs) specification using vitro models with temporally resolved single-cell transcriptomics and in-depth characterisation vivo datasets from nonhuman primates, including a 3D marmoset reference atlas. We elucidate molecular signature for transient gain competence fate peri-implantation epiblast development. Furthermore, show that both PGCs amnion arise transcriptionally similar TFAP2A-positive progenitors at posterior end embryo. Notably, genetic loss function experiments shows TFAP2A is crucial initiating PGC without detectably affecting subsequently replaced by TFAP2C as an essential component network fate. Accordingly, amniotic cells continue to emerge epiblast, but importantly, this also source nascent PGCs.

Language: Английский

---

Single-cell and single-nuclei RNA sequencing as powerful tools to decipher cellular heterogeneity and dysregulation in neurodegenerative diseases

Frontiers in Cell and Developmental Biology, Journal Year: 2022, Volume and Issue: 10

Published: Oct. 24, 2022

Neurodegenerative diseases affect millions of people worldwide and there are currently no cures. Two types common neurodegenerative Alzheimer’s (AD) Parkinson’s disease (PD). Single-cell single-nuclei RNA sequencing (scRNA-seq snRNA-seq) have become powerful tools to elucidate the inherent complexity dynamics central nervous system at cellular resolution. This technology has allowed identification cell states, providing new insights into susceptibilities molecular mechanisms underlying conditions. Exciting research using high throughput scRNA-seq snRNA-seq technologies study AD PD is emerging. Herein we review recent progress in understanding these state-of-the-art technologies. We discuss fundamental principles implications single-cell human brain. Moreover, some examples computational analytical required interpret extensive amount data generated from assays. conclude by highlighting challenges limitations application PD.

Language: Английский

---

Technology meets TILs: Deciphering T cell function in the -omics era

Cancer Cell, Journal Year: 2022, Volume and Issue: 41(1), P. 41 - 57

Published: Oct. 6, 2022

Language: Английский

---

Depth normalization for single-cell genomics count data

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2022, Volume and Issue: unknown

Published: May 6, 2022

Single-cell genomics analysis requires normalization of feature counts that stabilizes variance while accounting for variable cell sequencing depth. We discuss some the trade-offs present with current widely used methods, and analyze their performance on 526 single-cell RNA-seq datasets. The results lead us to recommend proportional fitting prior log transformation followed by an additional fitting.

Language: Английский

---

Neural manifold analysis of brain circuit dynamics in health and disease

Journal of Computational Neuroscience, Journal Year: 2022, Volume and Issue: 51(1), P. 1 - 21

Published: Dec. 16, 2022

Recent developments in experimental neuroscience make it possible to simultaneously record the activity of thousands neurons. However, development analysis approaches for such large-scale neural recordings have been slower than those applicable single-cell experiments. One approach that has gained recent popularity is manifold learning. This takes advantage fact often, even though datasets may be very high dimensional, dynamics tends traverse a much lower-dimensional space. The topological structures formed by these low-dimensional subspaces are referred as "neural manifolds", and potentially provide insight linking circuit with cognitive function behavioral performance. In this paper we review number linear non-linear learning, including principal component (PCA), multi-dimensional scaling (MDS), Isomap, locally embedding (LLE), Laplacian eigenmaps (LEM), t-SNE, uniform approximation projection (UMAP). We outline methods under common mathematical nomenclature, compare their advantages disadvantages respect use data analysis. apply them from published literature, comparing manifolds result application hippocampal place cells, motor cortical neurons during reaching task, prefrontal multi-behavior task. find many circumstances algorithms produce similar results methods, although particular cases where complexity greater, tend manifolds, at expense interpretability. demonstrate study neurological disorders through simulation mouse model Alzheimer's Disease, speculate help us understand circuit-level consequences molecular cellular neuropathology.

Language: Английский

---

Case-control analysis of single-cell RNA-seq studies

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2022, Volume and Issue: unknown

Published: March 18, 2022

Summary Single-cell RNA-seq (scRNA-seq) assays are being increasingly utilized to investigate specific hypotheses in both basic biology and clinically-applied studies. The design of most such studies can be often reduced a comparison between two or more groups samples, as disease cases healthy controls, treatment placebo. Comparative analysis scRNA-seq samples brings additional statistical considerations, currently there is lack tools address this common scenario. Based on our experience with comparative designs, here we present computational suite ( Cacoa – ca se- co ntrol nalysis ) carry out tests, exploration, visualization sample cohorts. Using multiple example datasets, demonstrate how application these techniques provide insights, avoid issues stemming from inter-individual variability, limited size, high dimensionality the data.

Language: Английский

---