Identifying tumor cells at the single-cell level using machine learning
40
Citation
67
Reference
10
Related Paper
Citation Trend
Abstract:
Abstract Tumors are complex tissues of cancerous cells surrounded by a heterogeneous cellular microenvironment with which they interact. Single-cell sequencing enables molecular characterization of single cells within the tumor. However, cell annotation—the assignment of cell type or cell state to each sequenced cell—is a challenge, especially identifying tumor cells within single-cell or spatial sequencing experiments. Here, we propose ikarus, a machine learning pipeline aimed at distinguishing tumor cells from normal cells at the single-cell level. We test ikarus on multiple single-cell datasets, showing that it achieves high sensitivity and specificity in multiple experimental contexts.Keywords:
Single-Cell Analysis
Cell type
Single cell sequencing
Abstract Although cell-in-cell structure was noted 100 years ago, the molecular mechanisms of ‘entering’ and the destination of cell-in-cell remain largely unclear. It takes place among the same type of cells (homotypic cell-in-cell) or different types of cells (heterotypic cell-in-cell). Cell-in-cell formation affects both effector cells and their host cells in multiple aspects, while cell-in-cell death is under more intensive investigation. Given that cell-in-cell has an important role in maintaining homeostasis, aberrant cell-in-cell process contributes to the etiopathology in humans. Indeed, cell-in-cell is observed in many pathological processes of human diseases. In this review, we intend to discuss the biological models of cell-in-cell structures under physiological and pathological status.
Cell type
Cell–cell interaction
Cell Signaling
Cite
Citations (34)
Cell Sorting
Single-Cell Analysis
Single cell sequencing
Cite
Citations (7)
Abstract Cell-free RNA (cfRNA) can be used to noninvasively measure dynamic and longitudinal physiological changes throughout the body. While there is considerable effort in the liquid biopsy field to determine disease tissue-of-origin, pathophysiology occurs at the cellular level. Here, we describe two approaches to identify cell type contributions to cfRNA. First we used Tabula Sapiens , a transcriptomic cell atlas of the human body to computationally deconvolve the cell-free transcriptome into a sum of cell type specific transcriptomes, thus revealing the spectrum of cell types readily detectable in the blood. Second, we used individual tissue transcriptomic cell atlases in combination with the Human Protein Atlas RNA consensus dataset to create cell type signature scores which can be used to infer the implicated cell types from cfRNA for a variety of diseases. Taken together, these results demonstrate that cfRNA reflects cellular contributions in health and disease from diverse cell types, potentially enabling determination of pathophysiological changes of many cell types from a single blood test.
Cell type
Cite
Citations (11)
ABSTRACT Hereditary diseases manifest clinically in certain tissues, however their affected cell types typically remain elusive. Single-cell expression studies showed that overexpression of disease-associated genes may point to the affected cell types. Here, we developed a method that infers disease-affected cell types from the preferential expression of disease-associated genes in cell types (PrEDiCT). We applied PrEDiCT to single-cell expression data of six human tissues, to infer the cell types affected in 1,113 hereditary diseases. Overall, we identified 110 cell types affected by 714 diseases. We corroborated our findings by literature text-mining and recapitulation in mouse corresponding tissues. Based on these findings, we explored features of disease-affected cell types and cell classes, highlighted cell types affected by mitochondrial diseases and heritable cancers, and identified diseases that perturb intercellular communication. This study expands our understanding of disease mechanisms and cellular vulnerability.
Cell type
Mendelian inheritance
Cite
Citations (0)
A radioactivity-based assay was developed to define the participation of radioactively labeled cell types within the milieu of unlabeled partners in multigeneric aggregates. The cell types in these multigeneric aggregations consisted of various combinations of 21 strains representing five genera of human oral bacteria. The coaggregation properties of each cell type, when paired individually with various strains, were delineated and were unchanged when the microbes took part in the more complex multigeneric aggregations. Competition between homologous labeled and unlabeled cells for binding to a partner cell type was achieved only when the homologous cells were mixed together before the addition of their partner cells. Attempts to displace a labeled cell type from an aggregate by subsequent addition of a large excess of the same unlabeled cell type were unsuccessful, which suggested that the forces that bound different cell types together were very strong and the cell-to-cell interactions were stable. However, a cell type that exhibited only lactose-reversible coaggregations with partners was easily and selectively released by the addition of lactose to multigeneric aggregates otherwise consisting solely of lactose-nonreversible cell-to-cell interactions. This not only indicates the independent nature of individual coaggregations but also suggests the involvement of lectinlike adhesins in these sugar-inhibitable coaggregations. Although the molecular mechanisms responsible for multigeneric aggregations are unknown, the principle of a common partner cell type serving as a bridge between two otherwise noncoaggregating cell types was firmly established by the observation of sequential addition of one cell type to another. Thus, competition, bridging, coaggregate stability, independent nature of interactions, and partner specificity are the key principles of adherence that form the framework for continued studies of multigeneric aggregates. While the human oral cavity is a prime example of a complex microbial community, collectively the community appears to consist of simple and testable individual interactions.
Cell type
Cell membrane
Cite
Citations (106)
Abstract Tumors are complex tissues of cancerous cells surrounded by a heterogeneous cellular microenvironment with which they interact. Single-cell sequencing enables molecular characterization of single cells within the tumor. However, cell annotation—the assignment of cell type or cell state to each sequenced cell—is a challenge, especially identifying tumor cells within single-cell or spatial sequencing experiments. Here, we propose ikarus, a machine learning pipeline aimed at distinguishing tumor cells from normal cells at the single-cell level. We test ikarus on multiple single-cell datasets, showing that it achieves high sensitivity and specificity in multiple experimental contexts.
Single-Cell Analysis
Cell type
Single cell sequencing
Cite
Citations (40)
The inherent heterogeneity of individual cells in cell populations plays significant roles in disease development and progression, which is critical for disease diagnosis and treatment. Substantial evidences show that the majority of traditional gene profiling methods mask the difference of individual cells. Single cell sequencing can provide data to characterize the inherent heterogeneity of individual cells, and reveal complex and rare cell populations. Different microfluidic technologies have emerged for single cell researches and become the frontiers and hot topics over the past decade. In this review article, we introduce the processes of single cell sequencing, and review the principles of microfluidics for single cell analysis. Also, we discuss the common high-throughput single cell sequencing technologies along with their advantages and disadvantages. Lastly, microfluidics applications in single cell sequencing technology for the diagnosis of cancers and immune system diseases are briefly illustrated.
Single cell sequencing
Single-Cell Analysis
Cite
Citations (84)
Heterogeneity of a cell population has been considered a fundamental property of cellular system. Single cell analysis is essential to understand the variation within a heterogenous cell population; however, most existing single cell analysis methods can only provide a glimpse of cell property at specific time point, unable to provide phenotypic information varying in time. Here we review single cell analysis assays we developed to monitor single cell behaviors and cell secretions over time. The assays also provide the capability of translocating cells to another substrate for downstream analysis of selected single cells. J Nat Sci, 5(1):e547, 2019
Single-Cell Analysis
Extracellular Vesicles
Cell type
Cite
Citations (0)
To understand the inhomogeneity of cells in biological systems, there is a growing demand on the capability of characterizing the properties of individual single cells. Since single-cell studies require continuous monitoring of the cell behaviors, an effective single-cell assay that can support time lapsed studies in a high throughput manner is desired. Most currently available single-cell technologies cannot provide proper environments to sustain cell growth and, proliferation of single cells and convenient, noninvasive tests of single-cell behaviors from molecular markers. Here, a highly versatile single-cell assay is presented that can accommodate different cellular types, enable easy and efficient single-cell loading and culturing, and be suitable for the study of effects of in vitro environmental factors in combination with drug screening. One salient feature of the assay is the noninvasive collection and surveying of single-cell secretions at different time points, producing unprecedented insight of single-cell behaviors based on the biomarker signals from individual cells under given perturbations. Above all, the acquired information is quantitative, for example, measured by the number of exosomes each single-cell secretes for a given time period. Therefore, our single-cell assay provides a convenient, low-cost, and enabling tool for quantitative, time lapsed studies of single-cell properties.
Exosome
Single-Cell Analysis
Cite
Citations (99)
Mendelian diseases tend to manifest clinically in certain tissues, yet their affected cell types typically remain elusive. Single-cell expression studies showed that overexpression of disease-associated genes may point to the affected cell types. Here, we developed a method that infers disease-affected cell types from the preferential expression of disease-associated genes in cell types (PrEDiCT). We applied PrEDiCT to single-cell expression data of six human tissues, to infer the cell types affected in Mendelian diseases. Overall, we inferred the likely affected cell types for 328 diseases. We corroborated our findings by literature text-mining, expert validation, and recapitulation in mouse corresponding tissues. Based on these findings, we explored characteristics of disease-affected cell types, showed that diseases manifesting in multiple tissues tend to affect similar cell types, and highlighted cases where gene functions could be used to refine inference. Together, these findings expand the molecular understanding of disease mechanisms and cellular vulnerability.
Cell type
Mendelian inheritance
Cite
Citations (2)